Initial Estimates of Mercury Emissions to the Atmosphere from Global

Apr 15, 2009 - During this particular time period, the largest mercury emissions are ... The average annual global burned area in GFEDv2 for the years...
0 downloads 0 Views 2MB Size
Environ. Sci. Technol. 2009, 43, 3507–3513

Initial Estimates of Mercury Emissions to the Atmosphere from Global Biomass Burning H . R . F R I E D L I , * ,† A . F . A R E L L A N O , † S. CINNIRELLA,‡ AND N. PIRRONE‡ National Center for Atmospheric Research, Boulder, Colorado 80307, and CNR-Institute for Atmospheric Pollution, Rende, Italy

Received September 23, 2008. Revised manuscript received February 26, 2009. Accepted March 16, 2009.

The average global annual mercury emission estimate from biomass burning (BMB) for 1997-2006 is 675 ( 240 Mg/year. This is equivalent to 8% of all currently known anthropogenic and natural mercury emissions. By season, the largest global emissions occur in August and September, the lowest during northern winters. The interannual variability is large and regionspecific, and responds to drought conditions. During this particular time period, the largest mercury emissions are from tropical and boreal Asia, followed by Africa and South America. They do not coincide with the largest carbon biomass burning emissions, which originate from Africa. Frequently burning grasslands in Africa and Australia, and agricultural waste burning globally, contribute relatively little to the mercury budget. The released mercury from BMB is eventually deposited locally and globally and contributes to the formation of toxic bioaccumulating methyl mercury. Furthermore, increasing temperature in boreal regions, where the largest soil mercury pools reside, is expected to exacerbate mercury emission because of more frequent, larger, and more intense fires.

1. Introduction Mercury is released to the atmosphere from many natural and anthropogenic sources. Among natural sources, mercury emissions from biomass burning have recently received growing attention due to their potentially significant contribution to the atmospheric mercury budget and due to their impact on global mercury pollution (1-6). The importance of mercury emissions to the atmosphere from biomass burning was first recognized for South America (7), probably as the result of the confluence of mercury pollution from artisan gold mining and ongoing clearing of tropical forests by burning for agricultural uses. Since 2000, research describing laboratory and field experiments extended to other geographic regions with extensive wildfire activity and provided basic understanding of the mercury release process during biomass burning (BMB). The concern for this newly recognized pathway of mercury is its toxicity and participation in the atmosphere-biosphere biogeochemical cycle, which includes conversion into methyl mercury, a bioaccumulating * Corresponding author phone: 303-497-1395; fax: 303-497-2920; e-mail: [email protected]. † National Center for Atmospheric Research. ‡ CNR-Institute for Atmospheric Pollution. 10.1021/es802703g CCC: $40.75

Published on Web 04/15/2009

 2009 American Chemical Society

compound hazardous to humans and other mammals, fish, birds, amphibians, reptiles, and insects. A few attempts have been made to estimate mercury emissions at different scales (1-6, 8). Traditionally, mercury emission estimates from BMB have been made from groundbased actuarial data of fuel consumptions by government agencies and mercury emission factors related to fuel consumption. Such methods suffer from several shortcomings: most of all the lack of measurements in regions where most BMB occurs, but also inconsistent assessment techniques. We are now in a period of rapid advances in satellite remote sensing, retrieval algorithms, and numerical modeling, providing essential data on fuel characterization, fire detection and burn area growth, fire intensity, and smoke plume composition and transport on a global scale, which make it possible to estimate global mercury emissions from biomass burning in a globally consistent manner (5, 6, 9). The present study combines outputs from a sophisticated global carbon emission model with mercury emission factors determined by best available experimental biome-specific data to arrive at global mercury emission estimates from biomass burning for the years 1997-2006. The advantage of this approach is its ability to produce a consistent, traceable assessment on a global scale at a time when only limited experimental data are available. It can serve as a starting point for modeling atmospheric mercury distribution and future assessments of global mercury source attribution.

2. Methods and Models The global mercury emission estimate is based on a satelliteconstrained bottom-up methodology. The output of a sophisticated carbon emission database for biomass burning (Global Fire Emission Database version 2 (GFEDv2)) was used to account for global fires in regions with similar fuel types and fire behavior (10). Mercury emission factors for different ecosystem types were retrieved by two methods: (i) groundbased measurements of the difference in mercury pools before and after a fire, and (ii) enhancement ratios (ER) of Hg and CO in plumes measured on the ground or by aircraft. Mercury emissions are then estimated as the product of carbon emissions and mercury emission factors. The GFEDv2 estimates of trace gas emissions are derived according to 14 geographic regions of the globe, generally exhibiting similar climate, vegetation, and fire dynamics (Figure 1). The database only considers three broad categories of fuel, i.e., extra tropical (temperate and boreal), nonforests (savanna, agricultural waste, and grass), and tropical forests (equatorial forests), consistent with the paucity of detailed biome-specific emission factors. We briefly describe in the following sections the carbon emission model used to generate GFEDv2 carbon emission

FIGURE 1. Map of regions used in GFEDv2. VOL. 43, NO. 10, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3507

estimates and our efforts to compile a reasonably comprehensive database for emission factors for mercury from BMB. 2.1. Carbon Emission Modeling. Burned Area. An estimate of burned area (BA) is available in GFEDv2 (10, 36) and is in general agreement with Canadian Interagency Forest Fire Centre (CIFFC) compilations and U.S. National Interagency Fire Center (NIFC) statistics. GFEDv2 tends to underestimate burned areas when compared to CIFFC (slope of ∼70%) and NIFC (slope of ∼83%), particularly for very large fires. The average annual global burned area in GFEDv2 for the years 1997-2006 is 332 ( 26 Mha year-1. The majority is due to burning in Africa (66%), followed by Australia (13%), Asia (10%), and South America (5%). On average, uncertainties in GFEDv2 global burned area estimates appear to be smaller (20-30%) than earlier estimates which had uncertainties over a factor of 2 (11). However, the differences in GFEDv2 burned area with other estimates can be large during some years and on local to regional scales (36). Fuel Loads. The available fuel load (FL) is defined in GFEDv2 as organic matter available for combustion and includes all above-ground herbaceous biomass, aboveground woody biomass, coarse woody debris, and litter. The Carnegie-Ames-Stanford Approach (CASA) biogeochemical model (12) with a spatial resolution of 1° × 1° and a temporal resolution of one month is used to estimate available fuel load and to simulate the carbon fluxes across the different pools of the terrestrial biosphere, including carbon losses from fires, herbivores, and fuel wood collection. The model uses the combination of satellite-derived estimates of incident photosynthetically active radiation (PAR) absorbed by the green plant canopy (fAPAR) and estimates of solar insolation for PAR to simulate net primary production (NPP). The average fuel load (dry matter) and its standard deviation, as estimated by GFEDv2 for the year 2000 across regions of similar vegetation types, ranges from 1.3 ( 1 kg m-2 in nonforests, to 18 ( 21 kg m-2 in tropical forests (mostly equatorial Asia), to 10 ( 8 kg m-2 in extra tropical forests (including boreal). These estimates are comparable on average to field measurements (13-15) and have uncertainties on a regional scale, of about a factor of 2. A large part of the uncertainty is attributed to poor representation of soil organic carbon and below-ground biomass, especially in boreal regions and peat lands, although GFEDv2 has paid special attention to this important question. Combustion Completeness. Combustion completeness (CC) is defined here as the ratio of fuel consumed from fires to total available fuels (based in dry mass). Measurements are associated with the types of fuel, fuel loads, and fuel configurations (including moisture) but may also vary significantly depending on fire practices, fire severity, and dynamics. The values range from about 0.2 for coarser fuels like stems and woody debris, which burn incompletely, to 1 (for complete combustion) for well-aired and dry litter (fine fuels). In GFEDv2, CC is allowed to vary in the biogeochemical model across the different carbon pools and from month to month. In addition, CC is increased in stems and coarse litter in areas with high levels of fire persistence. Uncertainties in CC can be attributed to the lack of direct observations on fire behavior across different ecosystems and throughout the burning season. Fuel Consumption. The amount of biomass burned (fuel consumed) is a product of burned area, fuel load, and combustion completeness. The release of carbon from biomass burning is calculated by assuming a carbon content of 45% of the dry biomass (16). 2.2. Mercury Emission Factors, EF(Hg)s. Emission factors, EF(Hg)s, are determined from either (i) airborne or ground-level measurements of smoke plume compositions (plume-based), or (ii) observed changes in the mercury pool in soils before and after a fire (soil-based). A convenient 3508

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 10, 2009

TABLE 1. Emission Factors Used for Different Biomes within Each of the 14 Regions Considered in This Studya region

extra tropical

nonforest

tropical

BONA TENA CEAM NHSA SHSA EURO MIDE NHAF SHAF BOAS CEAS SEAS EQAS AUST

315 242 242 242 242 242 41 242 242 315 242 242

41 41 41 41 41

198 198 198

a

242

41 41 41 41 41 41 41

198 198 198 315 198

In µg Hg/kg fuel.

method to assess mercury in fire plumes is to measure an enhancement ratio (ER), defined as ∆[Hg]/∆[CO], where ∆[Hg] is the sum of all Hg species in excess of background, and ∆[CO] is the difference between CO concentration in the plume and the background. For this work we have assumed that the relationship between EF and ER remains constant for all fires and are related by a factor of 1425. This value is based on the most robust aircraft measurements in fires in temperate/boreal forests (3), i.e., EF(Hg) ) (113 ( 59) µg/kg fuel burned and ER ) 0.793 × 10-7 ppbHg/ppbCO. ERs reported in the literature are presented in Table S1 (Supporting Information), and the average value is 1.54 × 10-7, which corresponds to an EF(Hg) of 220 µg Hg/kg fuel burned. The soil-based method is applicable to biomes where mercury resides predominantly in the organic soil/surface litter, i.e., to fires in temperate and boreal forests. The method is based on measuring the difference in the mercury pools in organic soil in adjacent plots before and after a fire (e.g., 20-23). In temperate and boreal forests, mercury resides >90% in the organic soil (e.g., in ref 24). Studies focusing on boreal or temperate forests suggest that the above-ground mercury fraction is small and can be neglected since it has been found to be less than 10% of the total Hg pool (21). Release of Hg from the underlying mineral layer is also considered negligible (21) although not universally so. The advantage of the soil-based method is that it only requires total mercury measurements in the organic soil/surface litter. However, sampling is subject to large short- and long-range pool variability. The available data for both methods were combined and grouped according to different landscapes. The means and corresponding ranges (min/max) are presented in Table S2 in the Supporting Information. The EF(Hg)s applied in the budget calculations are averages of all experimental data from Table S2 applied to specified biomes, or guesses based on analogies for biomes with no available experimental data. The emission factor choice is fraught with large uncertainties and variability of different types: e.g., the large interannual variability in equatorial Asia (EQAS) causing different fire severity (see section 3.3), or the paucity of measurements in the regions of largest carbon emissions such as Africa, equatorial Asia, and South America. The EF(Hg)s used for each category for each region are given in Table 1. Generally, the emission factors reflect the variation in fuel type, frequency, and/or severity of fires in the respective regions. EF(Hg)s for boreal regions such as BONA and BOAS represent large mercury pools with infrequent fire activities, whereas EF(Hg)s for savannas and grasslands (e.g., Africa,

TABLE 2. Regional and Global Emission Estimates for Mercury and Carbon (1997-2006) Hg emissions Mg Hg/year

a

carbon emissions Tg C/year

burned area Mha

effective EF µg Hg/kg fuel

regions

mean

SDa

mean

SD

mean

SD

mean

BONA TENA CEAM NHSA SHSA EURO MIDE NHAF SHAF BOAS CEAS SEAS EQAS AUST global borealb temperatec ROWd

22 6 22 13 95 2 0 83 58 99 7 57 192 19 675 121 9 545

16 3 25 10 39 1 0 13 7 83 2 35 216 9 240 85 3 224

42 16 56 38 294 14 1 618 571 170 47 144 276 133 2420 212 30 2178

30 6 61 25 97 5 0 74 68 124 12 84 312 35 382 128 7 360

2 2 3 4 12 2 0 141 78 9 17 12 4 46 332 11 4 316

1 0 2 1 2 1 0 13 9 5 5 5 4 18 26 5 1 26

233 178 175 157 145 72 17 60 46 263 67 177 312 65 279 248 89 134

SD for standard deviation.

b

BONA + BOAS. c TENA + EURO + MIDE.

SHSA, AUST) have frequent fire occurrences with smaller mercury pools. A particularly difficult assignment is for EQAS, where experimental data are unavailable and the interannual variability is extremely large.

3. Results The three primary results from this work are (i) the average annual emission estimates for the years 1997-2006 for each region and the composite globe, (ii) the seasonal variability, and (iii) the interannual variability for each region. This work presents an initial and globally consistent estimate of mercury emissions from biomass burning, based on best available regional to global carbon emission estimates and biomespecific mercury release characteristics. 3.1. Average Regional and Global Mercury Emissions from BMB (1997-2006). Using the procedure outlined above and carbon emission estimates from GFEDv2 as the basis for biomass burning activity, we estimate for global mercury emissions from biomass burning averaged across the period 1997-2006 to be 675 ( 240 Mg/year. The large range, which is calculated as the standard deviation across the time period, reflects the strong spatiotemporal variability of the emissions. Shown in Table 2 are the regional estimates for mercury and carbon for 1997-2006 averages. As can be seen later in section 3.3, the time frame of reporting is very important in the final tally because of the distinct interannual variability. The major emissions of mercury from BMB come from equatorial Asia (EQAS) (28%), boreal Asia (BOAS) (15%), and southern hemisphere South America (SHSA) (14%), and only in part from northern hemisphere Africa (NHAF) (12%), southern hemisphere Africa (SHAF) (9%), southeast Asia (SEAS) (8%), central America (CEAM) (4%), and Australia (AUST) (3%). Temperate North America (TENA) (1%), boreal North America (BONA) (3%), central Asia (CEAS), northern hemisphere South America (NHSA), Europe (EURO), and Middle East (MIDE) combined contribute little (2%) to the global budget. 3.2. Mean Seasonality for 1997-2006. We show in Figure 2 the progression of the mercury emission hot spots as they move seasonally across the globe. The mean seasonality for the period 1997-2006 is displayed in Table S3 in the Supporting Information. Globally, biomass burning occurs throughout the year, predominantly peaking in July-September and slightly

d

Rest of the world.

peaking again in December-February. A large fraction of this peak is due to fires in Africa during the dry season. In northern hemisphere Africa, burning occurs in the winter dry season within the savanna ecosystems (e.g., Sahara desert and central African rain forests). It begins in the Sahel in October and spreads south through November with highest burning activity in December and January. Fires in southern hemisphere Africa typically start in the woodlands of Zaire and Congo by early June and continue to peak across the southeast in the grasslands and shrub lands of Angola, Zambia, and Tanzania during August through September and Mozambique in October. A similar fire pattern occurs in South America, mostly in the cerrados of Brazil and along the arc of fire (or deforestation) in the Amazon rainforest, generally in August through September. Fires in tropical Asia (Indonesia, Malaysia, and the rest of Southeast Asia) are highly variable and are usually associated with very dry conditions, like El Nino events, facilitating the use of fire by land owners as a tool for land-clearing in the region. Because of the large peat deposits in Indonesia that are exposed during land-clearing, the carbon emissions in equatorial Asia are significantly higher during El Nino years (e.g., 1997-1998). Burning typically occurs in March-April over southeast Asia and in August-October in equatorial Asia. Fires in central America peak during April to June mainly due to deforestation and agriculture, while much of the bush fires in Australia occur in the shrublands in northern Australia and to some degree in southeast Australia during September through December. In the northern hemisphere temperate and boreal regions, a large fraction of fires are caused by lightning. Boreal forest fires generally take place in May to September when temperatures and lightning frequencies are high. Fires in boreal Asia typically start in May around Mongolia and spread north in Siberia during summer. Like equatorial Asia, fires in the boreal regions are also affected by droughts and very dry conditions. For example, fires in Alaska, Canada, and Siberia were especially high in 1997-1998 and 2004. Consequently, releases of carbon are high during these years also due to characteristically higher fuel loads (i.e., soil organic carbon in peat) in these regions. Biomass burning in western Russia and Europe is mostly associated with agriculture and generally occurs between spring and fall, while burning in the continental U.S. is predominantly from forest wild fires VOL. 43, NO. 10, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3509

FIGURE 2. Average monthly mercury emissions for the period 1997-2006. and prescribed burning, generally occurring from June to September. Note that the North American wildfires in recent years tend to be larger, start earlier, last longer, and be more intense (30). Carbon release from biomass burning in the continental U.S. is small relative to emissions from boreal forests in Alaska, Canada, and Siberia. Our estimates show a slightly different seasonal peak in mercury emissions, which are largest in August and March compared to carbon emissions that are largest in August and December. This is mainly due to the strong influence of high emission factors for these forested regions and low emission factors for fires in the savannas. This effect highlights the sensitivity of the emissions to estimates of burned area and emission factors. Furthermore, this result indicates that while uncertainties on estimates of carbon emissions from fires in savannas (as well as burning from agricultural waste) play an important role, they are less significant with regards to mercury. 3.3. Interannual Variability of Mercury Emissions. Shown in Figure 3 are annual emissions for different regions during the 10-year period covered by this report. There is a large interannual variability of mercury emissions across the different regions, particularly in EQAS, SHSA, BOAS, BONA, and CEAM. There were large amounts of mercury released during the strong El Nino year of 1997-1998 and during drought conditions in 2003-2004. The largest interannual variability occurred in Indonesia where peat deposits were available as fuel loads, as well as in boreal region and in 3510

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 10, 2009

FIGURE 3. Annual mercury emissions for 1997-2006 (see Figure 1 for description of regions). deforestation regions of tropical America. This impact can be seen in fires in boreal North America, where large amounts of mercury were released during the extreme fire seasons of 1997-1998 and 2003-2004, relative to nonmajor fire years such as during 2000 (26). This is consistent with observations of trace gases in smoke plumes such as CO, where it was

observed that fires in Alaska contributed to poorer air quality in continental U.S (e.g., 31). On the other hand, there is a more or less uniform contribution of mercury emissions to the global budget from Africa for the 10-year period. The difference in interannual variability is also reflected in the standard deviation of the mean estimates discussed in Table 2 and is consistent with previous studies on biomass burning (e.g., 32, 10). Indeed, there is a clear indication of a strong relationship between biomass burning, precipitation, and temperature in conjunction with fire practices in different regions.

4. Discussion The data confirm what would be logically expected: fires occur predominantly where most fuels are located, and most mercury is released where the largest mercury pools reside. The climate and geography of a specific region primarily determine the distribution and type of vegetation (see cdiac.orln.gov) and hence the amount of fuel available for burning. Mercury input to the biosphere is mostly by wet and dry deposition of atmospheric mercury, the rate of which varies by a factor of about 2 over vegetated land areas away from point sources (33). Mercury resides partially in aboveground plant structures when no or small amounts of organic soil is present, but in areas with substantial organic soil, most mercury resides underground associated with carbon, and chemically bound with reduced sulfur. The size of the mercury (and carbon) pool depends on the length of accumulation between fire events and the prevailing climate. Warm and wet climates are characterized by rapid carbon uptake in vegetation, but also by rapid decomposition upon senescence. The result is little organic soil formation and carbon and mercury sequestration. For example, African savannas accumulate only 0-2.5 cm of organic soil (15). By contrast, in cold and wet climates, large pools of carbon and mercury are accumulated in boreal forests, peat, and permafrost (26), and across the globe, gradients between the two extremes exist. In addition to the climatic and biological constraints, the fire frequency determines steady-state pool size of mercury and carbon. Savannas burn frequently (intentionally or by accident), typically annually or biannually, while boreal forest burns at 50-200 year time scales, and wet tropical forests rarely burn at all. In areas of active deforestation, fuel is intentionally and completely combusted. Remote sensing and carbon emission models can approximately represent these diverse conditions. It becomes obvious that EF(Hg) should also be responsive to the considerations discussed above for carbon emissions. To fulfill such criteria at this time is undeliverable, and we have to resort to the best available data (see section 2.2). The key finding of this work is the dominant impact of fires in the boreal region and in the tropical forests of Asia and South America on mercury emissions across the globe. While the majority of carbon emissions can be attributed to fires in African savannas, the mercury emissions show major contributions from equatorial Asia, boreal Asia, and southern hemisphere South America. Another key observation is the strikingly large interannual variability under some climatic circumstances, exemplified by the overwhelming emissions during the Indonesian peat fires in 1997-1998. Including years with extreme fire in budget estimates can skew the averaged data, thereby highlighting the need for explicit statements about the time frame covered, when emission budgets are reported. Future emissions will likely change. While regions with frequent BMB emit recently deposited (and modified by biogeochemical cycling) mercury, regions with large pools become impacted by more frequent burning and release of pulses of mercury under conditions of global or regional warming (30).

For comparison purposes, we estimated global mercury emission based on reported carbon monoxide bottom-up and top-down emission inventories from BMB and using a global average emission factor for mercury of 220 µg Hg/kg fuel (see section 2.2). Estimates range from 708 to 1346 Mg Hg/year, higher than the 675 ( 240 Mg/year estimates derived from the GFEDv2 model and biome-specific EF(Hg). While this approach considers only one global emission factor for mercury along with global CO emission estimates from different sources (which oversimplifies the spatiotemporal heterogeneity), this comparison serves as a confirmation of global mercury emission from BMB. A comparison of published values (not shown) for regional (country) carbon and mercury emission from biomass burning with values from this work shows significant differences in some cases; some are explainable by differences in carbon emission estimates and use of different EF(Hg) values, and still others reflect different model assumptions.

5. Impact and Implications Our global estimate of 675 ( 240 Mg/year (average 1997-2006) accounts for about 8% of all currently known anthropogenic and natural atmospheric emissions (9). While this contribution may seem minor on a global scale, local and regional impacts can be much more pronounced. In assessing health impacts, the main question is the location of deposition and subsequent availability for conversion into methyl mercury. BMB mercury is emitted in the form of gaseous elemental mercury (GEM) with a lifetime of about one year (but variable) and particulate mercury (pHg) associated with smoke particles with a lifetime of days to weeks. The speciation of mercury is dependent on fuel conditions (e.g., moisture content) and fire dynamics: pHg fractions between 4% and 13% have been observed in temperate and boreal wildfires (3, 18), which result in local or regional deposition. GEM, the major mercury species in BMB plumes, becomes incorporated into the global atmospheric pool. The injection height is a major factor for longrange transport: the major sources of mercury from fires occur in regions where transport plays an important role in the distribution of BMB plumes and atmospheric mercury across the globe. In particular, mercury released from fires in equatorial Asia and tropical South America are often transported to higher altitudes due to strong convection in these regions and can then be dispersed efficiently over a large area. In a similar manner, mercury released in boreal regions can be injected to the lower stratosphere (34), thereby influencing the distribution of mercury in the northern hemisphere middle to high latitudes. The fate and transport of emitted mercury is difficult to define because of the regionally different injection heights which lead to unique plume trajectories and associated chemistries and deposition. New studies on mercury pools and Hg emissions from the major BMB regions, e.g., Africa, southeast Asia, or Siberia, are mandatory. This work could also serve as part of the emissioninventoryforglobalmodelingofmercurydistribution. Mercury from BMB is clearly a global pollutant, and its control should be encouraged. Since mercury in vegetation and organic soils originates largely from the deposition from the global atmospheric pool, restricting the global release of anthropogenic mercury would over time reduce the atmospheric and vegetation/soil pools and thus the release potential in the case of fires. The release of mercury from biomass burning is partially under human control. Limiting the burning of tropical and boreal forests (EQAS, SHSA, BOAS) would have two beneficial effects: reducing the source of mercury releases to the atmosphere during burning, and maintaining a sink for atmospheric mercury in the vegetation and organic soil. VOL. 43, NO. 10, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3511

Warming as a result of climate change will be felt particularly in boreal forests (35), which harbor large carbon and mercury pools, and are expected to experience more frequent, larger, and more severe wildfires in the future.

Acknowledgments We thank James T. Randerson of University of California, Irvine, Guido R. van der Werf of Vrije Universiteit Amsterdam, Louis Giglio of Science Systems and Applications, Inc., Maryland, G. James Collatz of NASA Goddard Space Flight Center, Maryland, and Prasad S. Kasibhatla of Duke University for GFEDv2 emission data. We also thank C. Wiedinmyer and G. Pfister for helpful comments. H.F. is funded by the National Center for Atmospheric Research, which is sponsored by the National Science Foundation. A.F.A. acknowledges NASA support under Grant NNX07AL57G while N.P. and S.C. are supported by the Italian Ministry of Environment.

(15)

(16)

(17)

(18)

(19)

Supporting Information Available Table for observed molar enhancement ratios and a table for emission factors used in the emission calculations, as mentioned in section 2.2. Summary table for the mean seasonality of Hg emissions, as described in section 3.2. This material is available free of charge via the Internet at http:// pubs.acs.org.

(20)

(21)

Literature Cited (1) Brunke, E. G.; Labyschagne, C.; Slemr, F. Gaseous mercury emissions from a fire in the Cape Peninsula, South Africa, during January 2000. Geophys. Res. Lett. 2001, 28, 1483–1486. (2) Friedli, H. R.; Radke, L. F.; Lu, J. Y. Mercury in smoke from biomass fires. Geophys. Res. Lett. 2001, 28 (17), 3223–3226. (3) Friedli, H. R.; Radke, L. F.; Prescott, R.; Hobbs, P. V.; Sinha, P. Mercury emissions from the August 2001 wildfires in Washington State and an agricultural waste fire in Oregon, and atmospheric mercury budget estimates. Global Biogeochem. Cycles 2003, 17 (2), 1039. (4) Sigler, J. M.; Lee, X.; Munger, W. Emission and long-range transport of gaseous mercury from a large-scale Canadian boreal forest fire. Environ. Sci. Technol. 2003, 37, 4343–4347. (5) Cinnirella, S.; Pirrone, N. Spatial and temporal distribution of mercury emissions from forest fires in Mediterranean region and Russian Federation. Atmos. Environ. 2006, 40, 7346–7361. (6) Cinnirella, S.; Pirrone, N.; Allegrini, A.; Guglietta, D. Modeling mercury emissions from forest fires in the Mediterranean region. Environ. Fluid Mech. 2008, 8, 129–145. (7) Veiga, M. M.; Meech, J. A.; Ornante, N. Mercury pollution from deforestation. Nature (London) 1994, 368, 816–817. (8) Wiedinmyer, C.; Friedli, H. Mercury emission estimates from fires: An initial inventory for the United States. Environ. Sci. Technol. 2007, 41, 8092–8098. (9) Friedli, H. R.; Arellano, A. F., Jr.; Cinnirella, S.; Pirrone, N. Mercury emissions from global biomass burning: spatial and temporal distribution. In Mercury Fate and Transport in the Global Atmosphere: Measurements, Models and Policy Implications; Pirrone, N., Mason R., Eds.; UNEP, 2008. (10) van der Werf, G. R.; Randerson, J. T.; Giglio, L.; Collatz, G. J.; Kasibhatla, P. S.; Arellano, A. Interannual variability in biomass burning emissions from 1997 to 2004. Atmos. Chem. Phys. 2006, 6, 3423–3441. (11) Kasischke, E. S.; Penner, J. E. Improving global emissions of atmospheric emissions form biomass burning. J. Geophys. Res. 2004, 109, D14S01,; doi10.1029/2004JD004972. (12) Potter, C. S.; Randerson, J. T.; Field, C. B.; Matson, P. A.; Vitousek, P. M.; Mooney, H. A.; Klooster, S. A. Terrestial ecosystem productionsA process model based on global satellite and surface data. Global Biogeochem. Cycles 1993, 7 (4), 811–841. (13) Guild, L. S.; Kauffman, J. B.; Ellingson, L. J.; Cummings, D. L.; Castro, E. A.; Babbitt, R. E.; Ward, D. E. Dynamics associated with total aboveground biomass, C, nutrient pools, and biomass burning of primary forest and pasture in Rondonia, Brazil during SCAR-B. J. Geophys. Res. 1998, 103 (D24), 32091–32100. (14) Hobbs, P. V.; Reid, J. S.; Herring, J. A.; Nance, J. D.; Weiss, R. E.; Ross, J. L.; Hegg, D. A.; Ottmar, R. D.; Liousse, C. Particles and trace gas measurements in smoke from prescribed burns of 3512

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 10, 2009

(22)

(23)

(24)

(25)

(26)

(27)

(28)

(29) (30)

(31)

(32)

(33)

(34)

forest products in the Pacific Northwest. In Biomass Burning and Global Change; Levine, J. S., Ed.; MIT Press: Cambridge, MA, 1996; Vol. 1, pp 697-715. Shea, R. W.; Shea, B. W.; Kauffman, J. B.; Ward, D. E.; Haskins, C. I.; Scholes, M. C. Fuel biomass and combustion factors associated with fires in savanna ecosystems in South Africa and Zambia. J. Geophys. Res. 1996, 101 (D19), 23551–23568. Andreae, M. O.; Merlet, P. Emission of trace gases and aerosols from biomass burning. Global Biogeochem. Cycles 2001, 15 (4), 955–966. Weiss-Penzias, P.; Jaffe, D.; Swartzendruber, P.; Hafner, W.; Chand, D.; Prestbo, E. Quantifying Asian and biomass burning sources of mercury using the Hg/CO ratio in pollution plumes observed at the Mount Bachelor Observatory. Atmos. Environ. 2007, 41, 4366–4379. Friedli, H. R.; Radke, L. F.; Lu, J. Y.; Banic, C. M.; Leaitch, W. R.; MacPherson, J. I. Mercury emissions from burning of biomass from temperate North American forests: Laboratory and airborne measurements. Atmos. Environ. 2003, 37, 253–267. Ebinghaus, R.; Slemr, F.; Brenninkmeijer, C. A. M.; van Velthoven, P.; Zahn, A.; Hermann, M.; Sullivan, D. A.; Oram, D. E. Emission of gaseous mercury from biomass burning in South America in 2005 observed during CARIBIC flights. Geophys. Res. Lett. 2007, 34, L08813; doi 10.1029/2006GL028866. Harden, J. W.; Neff, J. C.; Sandberg, D. V.; Turetsky, M. R.; Ottmar, R.; Gleixner, G.; Fries, T. L.; Manies, K. L. Chemistry of burning the forest floor during the FROSTFIRE experimental burn, interior Alaska,1999. Global Biogeochem. Cycles 2004, 18, GB3014; doi 10.1029/2003GB002194. Engle, M. A.; Sexauer Gustin, M.; Johnson, D. W.; Murphy, J. F.; Miller, W. W.; Walker, R. F.; Wright, J.; Markee, M. Mercury distribution in two Sierran forest and one desert sagebrush steppe ecosystems and the effects of fire. Sci. Total Environ. 2006, 367, 222–233. Biswas, A.; Blum, J.; Keeler, J. A. Comparison of methods to estimate mercury emissions during wildfire. 8th International Congress on Mercury as a Global Pollutant, 2006, T-125. Biswas, A.; Blum, J.; Keeler, J. Mercury storage in a central Washington forest and release during the 2001 Rex Creek Fire. STOTEN-D-07-011986, 2008, in review. Biswas, A.; Blum, J. D.; Klaue, B.; Keeler, J. G. Release of mercury from Rocky Mountain forest fires. Global Biogeochem. Cycles 2007, 21, GB1002; doi 10.1029/2006GB002696. Friedli, H. R.; Radke, L. F.; Payne, N. J.; McRae, D. J.; Lynham, T. J.; Blake, T. W. Mercury in vegetation and organic soil at an upland boreal forest site in Prince Albert National Park, Saskatchewan, Canada. J. Geophys. Res. 2007, 112, G01004; doi 10.1029/2005JG000061. Turetsky, M. R.; Harden, J. W.; Friedli, H. R.; Flannigan, M.; Payne, N.; Crook, J.; Radke, L. F. Wildfires threaten mercury stocks in northern soils. Geophys. Res. Lett. 2006, 33, L16403; doi 10.1029/2005GL025595. Woodruff, L. G.; Harden, J. W.; Cannon, W. F.; Gough, L. P. Mercury loss from the forest floor during wildland fire. American Geophysical Union, 2001, fall meeting, abstract B32B-0117. Obrist, D.; Moosmueller, H.; Schuermann, R.; Antony Chen, L.-W.; Kreidenweis, S. M. Particulate phase and gaseous elementary mercury speciation in biomass combustion: Controlling factors and correlation with particulate matter emission. Environ. Sci. Technol. 2008, 721–727. Friedli, H. R. et al. Mercury emissions from experimental burns of Southern African vegetation. Unpublished manusript. Westerling, A. L.; Swetnam, T. W. Interannual to decadal drought and wildfire in the Western United States. EOS 2003, 84 (49), 545–560. Pfister, G.; Emmons, L. K.; Hess, P. G.; Honrath, R. E.; Lamarque, J.-F.; ValMartin, M.; Owen, R. C.; Avery, M.; Browell, E.; Holloway, J.; Nedelec, P.; Purvis, R.; Ryerson, T.; Sachse, G.; Schlager, H. Ozone Production from the 2004 North American Boreal Fires. J. Geophys. Res. 2006, 111, D24S07. Duncan, B. N.; Martin, R. V.; Staudt, A. C.; Yevich, R.; Logan, J. L. Interannual and seasonal variability of biomass burning emissions constrained by satellite observations. J. Geophys. Res. 2003, 108 (D2), 4100. doi:10.1029/2002JD002378. Seigneur, C.; Karamchandani, P.; Lohman, K.; Vijayaraghvan, K.; Shia, R. Multiscale modeling of the atmospheric fate and transport of mercury. J. Geophys. Res. 2001, 106, 27795–27809. Fromm, M.; Alfred, J.; Hoppel, K.; Hornstein, J.; Bevilacqua, B.; Shettle, E.; Servrandcx, R.; Li, Z.; Stocks, B. J. Observation of boreal fire smoke in the stratosphere by POAM III, SAGE II and Lidar in 1998. Geophys. Res. Lett. 2000, 27, 1407–1410.

(35) Randerson, J. T.; Liu, H.; Flanner, M. G.; Chambers, S. G.; Jin, Y.; Hess, P. G.; Pfister, G.; Mack, M. C.; Treseder, K. K.; Welp, L. R.; Chapin, F. S.; Harden, J. W.; Goulden, M. L.; Lyons, E.; Neff, J. C.; Schuur, E. A. G.; Zender, C.S. The impact of boreal forest fires on climate warming. Science 2006, 314, 1130– 1132.

(36) Giglio, L.; van der Werf, G. R.; Randerson, J. T.; Collatz, G. J.; Kasibhatla, P. S. Global estimation of burned area using MODIS active fire observations. Atmos. Chem. Phys. 2006, 6, 957–974.

ES802703G

VOL. 43, NO. 10, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3513