Fate of Elemental Mercury in the Arctic during ... - ACS Publications

Mar 9, 2004 - NIELS Z. HEIDAM, BJARNE JENSEN,. PETER W Å HLIN, AND. GERALD GEERNAERT †. National Environmental Research Institute,...
0 downloads 0 Views 1021KB Size
Environ. Sci. Technol. 2004, 38, 2373-2382

Fate of Elemental Mercury in the Arctic during Atmospheric Mercury Depletion Episodes and the Load of Atmospheric Mercury to the Arctic HENRIK SKOV,* JESPER H. CHRISTENSEN, MICHAEL E. GOODSITE, NIELS Z. HEIDAM, BJARNE JENSEN, PETER WÅHLIN, AND GERALD GEERNAERT† National Environmental Research Institute, Frederiksborgvej 399, 4000 Roskilde, Denmark

Atmospheric mercury depletion episodes (AMDEs) were studied at Station Nord, Northeast Greenland, 81°36′ N, 16°40′ W, during the Arctic Spring. Gaseous elemental mercury (GEM) and ozone were measured starting from 1998 and 1999, respectively, until August 2002. GEM was measured with a TEKRAN 2735A automatic mercury analyzer based on preconcentration of mercury on a gold trap followed by detection using fluorescence spectroscopy. Ozone was measured by UV absorption. A scatter plot of GEM and ozone concentrations confirmed that also at Station Nord GEM and ozone are linearly correlated during AMDEs. The relationship between ozone and GEM is further investigated in this paper using basic reaction kinetics (i.e., Cl, ClO, Br, and BrO have been suggested as reactants for GEM). The analyses in this paper show that GEM in the Arctic troposphere most probably reacts with Br. On the basis of the experimental results of this paper and results from the literature, a simple parametrization for AMDE was included into the Danish Eulerian Hemispheric Model (DEHM). In the model, GEM is converted linearly to reactive gaseous mercury (RGM) over sea ice with temperature below -4 °C with a lifetime of 3 or 10 h. The new AMDE parametrization was used together with the general parametrization of mercury chemistry [Petersen, G.; Munthe, J.; Pleijel, K.; Bloxam, R.; Vinod Kumar, A. Atmos. Environ. 1998, 32, 829-843]. The obtained model results were compared with measurements of GEM at Station Nord. There was good agreement between the start and general features periods with AMDEs, although the model could not reproduce the fast concentration changes, and the correlation between modeled and measured values decreased from 2000 to 2001 and further in 2002. The modeled RGM concentrations over the Arctic in 2000 were found to agree well with the temporal and geographical variability of the boundary column of monthly average BrO observed by the GOME satellite. Scenario calculations were performed with and without AMDEs. For the area north of the Polar Circle, the mercury deposition increases from 89 tons/year for calculations without an AMDE to * Corresponding author phone: +4546301162; fax: +4546301214; e-mail: [email protected]. † Current address: Institute of Geophysics and Planetary Physics, Los Alamos National Laboratory, MS C-305, Los Alamos, NM 87545. 10.1021/es030080h CCC: $27.50 Published on Web 03/09/2004

 2004 American Chemical Society

208 tons/year with the AMDE. The 208 tons/year represent an upper limit for the mercury load to the Arctic.

Introduction Mercury is found at high levels in marine animals at many places in the Arctic and North Atlantic Ocean (1). The present levels of mercury in sea animals have a negative effect on the health of the local populations who use these animals as food supply (2). Furthermore, the input of atmospheric mercury to the Arctic environment has at least tripled as compared to preindustrial time (3). The lifetime of gaseous elemental mercury (GEM) in the atmosphere (95% of atmospheric mercury) is in general about 1 year (4). However, in the Arctic during spring, the lifetime of GEM is significantly shorter, and GEM is depleted in less than 1 day during mercury depletion episodes (AMDEs) (57). During an AMDE, GEM is converted to oxidized mercury in the gas phase, the so-called reactive gaseous mercury (RGM) that deposits quickly into the ground (6, 8). AMDE occurs typically in the Arctic from March to June, which coincidentally occurs when the marine arctic ecosystems are also extremely active due to the increasing solar flux combined with the melting of sea ice (9). Therefore, it is hypothesised that there is a higher efficiency of bioaccumulation of mercury than would be expected from extrapolating data from mid-latitudes to the Arctic. Thus, it is very important to get a full understanding of the processes responsible for the AMDEs. Ozone has been observed to be depleted during the Arctic spring since 1986 (10), and it is well-accepted that it is due to photochemical degradation. More recently, it has been demonstrated that GEM is depleted as well and that GEM is strongly correlated with ozone during an AMDE (5). Solar radiation and surface temperature of marine ice either control or are proxies for processes driving the depletion of GEM (6). However, little is known about the reaction(s) transforming GEM to RGM. Several hypotheses have been proposed where Cl and/or Br atoms or ClO and/or BrO are common candidates initiated by the heterogeneous reaction between ozone and sea salt chloride or bromide on the surface of sea ice with temperatures below -4 °C. Thus, AMDE is limited to areas exposed to marine air. Alhough the ratio is 300:1 of Cl-/Br- in seawater, a series of physical and chemical processes favors the liberation of bromine as compared to chlorine from the sea ice (11). Measurements of GEM and ozone at Station Nord, Northeast Greenland are presented in this paper. The data are treated using basic reaction kinetic theory and physical theory for mixing of gases in the atmosphere. The results of the measurements presented here and results from other studies are used to make a simple parametrization of AMDE, which is then used in DEHM (12, 13). The results of the model calculations are compared with measured values of GEM and with satellite observations of BrO (from the GOME satellite). Finally, the importance of AMDE for the burden of atmospheric mercury to the Arctic is determined by scenario calculations with and without AMDE.

Experimental Procedures Measurements. The monitoring site is at the Danish military base at Station Nord, Northeast Greenland located at 81°36′N, 16°40′W; see Figure 1a. The measurements were performed at the Danish AMAP site, Flyger’s Hytte, a laboratory hut located approximately 3 km south of the central complex of VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2373

FIGURE 1. (a) Greenland with the location of Station Nord. (b) The position of the air monitoring site Flyger’s Hytte at Station Nord. buildings, as shown on the map in the Figure 1b. The temperature in the hut was constant at 15 ( 1 °C. Weekly average concentrations of atmospheric bromine were determined from samples collected at 40 L/min through a 42 mm inlet diameter on a particle filter, the first of a series of 47 mm filters in a filter pack. A detailed description of the filter pack system is given elsewhere (14). The resulting filters were transported to the laboratory where they were analyzed by proton induced X-ray emission (PIXE) that is capable of detecting aluminum and all heavier elements (with the exception of mercury). The results of the analysis of bromine showed that the concentration of bromine atoms was the same on the front and reverse side of the filter. If bromine were present only in particle phase, it would have been observed only on the front side. Therefore, a significant fraction of the measured bromine must have been present in the gas phase (15). The measured bromine is thus called filterable bromine, fBr. The uncertainty of the method is estimated to be 25% at a 95% confidence interval. Ozone and GEM were measured from 1998 and 1999, respectively, and until June 2002. GEM was measured only from February through mid-summer (ending at July or August) each year. Ozone was measured with an UV absorption monitor, API, with a detection limit of 1 ppbv and an uncertainty of 3% for concentrations above 10 ppbv and 6% for concentrations below 10 ppbv (all uncertainties are at a 95% confidence interval) (16). GEM was measured by a TEKRAN 2537A mercury analyzer. The principle of the instrument is as follows: a measured volume of sample air is drawn through a gold trap that quantitatively retains elemental mercury. The collected mercury is desorbed from the gold trap by heat and is transferred by argon into the detection chamber, where the amount of mercury is detected by cold vapor atomic fluorescence spectroscopy. The detection limit is 0.1 ng/m3, and the reproducibility for concentrations above 0.5 ng/m3 is within 20% (at a 95% confidence interval) based on parallel measurements with two TEKRAN 2537A mercury analyzers. It is not at present reasonable to give the combined 2374

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004

uncertainty of the method following the guidelines of ISO 14956, as the exact identity of the measured mercury is unknown, although GEM is determined as the dominant compound (17). To protect the instrument against humidity and sea salt, a soda-lime trap was placed in the sample line just in front of the analyzer before the 2001 season to avoid passivation of the gold traps (18), and a heated sample line was used as well. However, no change in the level of GEM at Station Nord was observed after the installation of the trap and heated line. Parallel measurements of GEM in Denmark at a site not directly influenced by sea spray with and without soda-lime trap showed agreement within the experimental uncertainty. Shortly after the start in 2002, there were problems with the TEKRAN instrument due to a decrease in the voltage, and it did not work correctly before April 20, 2002. Model. The Danish Eulerian Hemispheric Model System (DEHM) is described in detail elsewhere (12, 13), so only a short description is given here. The model system consists of two parts: a meteorological part based on the PSU/NCAR Mesoscale Model version 5 (MM5) (19) and an air pollution model part, the DEHM model. The model system is driven by global meteorological data obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) on a 2.5 × 2.5° grid with a time resolution of 12 h. The DEHM model is based on a set of coupled full 3-D advection-diffusion equations, one equation for each species. The horizontal mother domain of the model is defined on a regular 96 × 96 grid that covers most of the Northern Hemisphere with a grid resolution of 150 × 150 km at 60 °N. The vertical resolution is defined on an irregular grid with 20 layers up to about 15 km reflecting the structure of the atmosphere. The chemistry scheme of mercury in the atmosphere outside the Arctic region includes 13 mercury species: three in the gas phase (Hg0, HgO, and HgCl2), nine species in the aqueous phase, and one in particulate phase and is adopted from the literature (20). Within the Arctic region, an additional first-order reaction of GEM was added to the chemical scheme where GEM reacts to form a RGM with properties determined

FIGURE 2. Hourly ozone mixing ratios and weekly concentration of fBr measured from 1999 to 2002 at Station Nord, Northeast Greenland. GEM is measured in the period from September 25, 1999 to August 23, 2000; February 14, 2001 to August 23, 2001; and April 26, 2002 to June 29, 2002.

experimentally in the Arctic; see following description. The reaction takes place inside the boundary layer over sea ice during sunny conditions to mimic an AMDE in accordance with ref 6. The reaction rate constant for the first-order removal is based on the observed removal rates of mercury at Station Nord. The measurements gave a first-order lifetime of GEM between 3 and 10 h, so scenario calculations with this range of lifetime were performed. The fast first-order oxidation stopped when the surface temperature exceeded -4 °C. This temperature appears to be crucial for the presence of AMDE as Br2 and BrCl are formed at the surfaces of refreezing leads (6). Except for the AMDEs, the fate of GEM is controlled by a slow chemical removal in the gas phase and uptake by cloudwater. The present version of the chemical scheme is the first step toward a chemical scheme, which simulates the chemical processes during AMDEs. Currently, a new scheme with a more comprehensive treatment of the chemical reactions is under development for the next generation of the model. The dry deposition velocities of the RGM species formed during AMDEs are based on the resistance method, where the surface resistance is similar to the one for nitric acid (i.e., 0 s m-1). This value is based on direct flux measurements of RGM (6, 8). It is important to stress that RGM formed during AMDEs most probably is not equal to HgO (see the discussion of the GEM reactions). The wet deposition of reactive and particulate mercury is parametrized by using a simple scavenging coefficients formulation with different in- and below-cloud scavenging coefficients (12). The emissions of anthropogenic mercury are based on the global inventory of mercury emissions for 1995 on a 1 × 1° grid (21) including emissions of GEM, RGM, and total particulate mercury (TPM). The model does not contain any reemissions from land and oceans. Instead, a background concentration of 1.5 ng/m3 of Hg0 is used as initial concentrations and boundary conditions. The mercury model has been run for the period of October 1998 to October 2002.

Results and Discussion Measurements. Figure 2 shows the results of ozone and GEM measurements together with concentrations of fBr. Ozone and GEM were rather stable from September/October until

the end of February/beginning of March. Then, a highly perturbed period appeared where both ozone and GEM were depleted to zero from, respectively, about 40 ppbv and 1.5 ng/m3. The concentrations remain at zero for periods that may last some hours up to several days before suddenly rising again. At the same time. fBr increases and reaches a maximum of about 10 ng/m3. In July, the ozone concentration stabilizes just above 20 ppbv, and then it slowly increases to about 40 ppbv in September/October, and fBr decreases to values close to zero. GEM was measured from February to the end of July or to the beginning of August. The measurements focused on the description of the AMDEs. Previously, ozone and GEM have been observed to deplete simultaneously and to be highly correlated during AMDEs (5). This is indeed confirmed by the present data set; see Figure 3. After the depletion period, some very high concentrations of GEM appeared with values above 2 ng/m3 in 2000 and at a maximum of 5.7 ng/m3 in 2002. In 2001, concentrations of up to 1.9 ng/m3 appeared. High values after AMDEs are also observed at Alert (5), Barrow (6), and Svalbard (7), and they are attributed to reemission of mercury to the atmosphere. The strong correlation between ozone and GEM suggests that they are dependent on a mutual factor. A direct reaction between ozone and GEM can be ruled out due to the long lifetime of GEM with respect to the present ozone concentration (4). In a field study (22), BrO was observed to build up when ozone is decreased due to the reaction

O3 + Br f O2 + BrO

(1)

Therefore, serious candidates for GEM removal are Br or BrO. However, Cl and ClO cannot be ignored, as significant Cl removal of organic compounds has been observed during AMDEs (e.g., ref 23), and the importance of these species depends on their concentrations and their reactivity toward GEM. The lifetime of GEM is observed to be typically about 10 h during AMDEs. Up to 30 ppt of ClO and 30 ppt BrO have been observed (24); thus, the resulting rate constants for the reactions between GEM and BrO and/or ClO can be estimated to be in the order of 4 × 10-14 cm3 molecule-1 s-1; see reactions VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2375

FIGURE 3. GEM against ozone concentrations at Station Nord, Northeast Greenland including a regression line obtained by orthogonal regression analysis. Data were selected from Figure 2 where at least three consecutive concentrations were decreasing on both ozone and GEM and where the initial GEM concentration was above 0.4 ng/m3. Only data from 2000 and 2001 were used as high concentrations in 2002 indicate the presence of other processes than in 2000 and 2001. 2 and 3, respectively

Hg + BrO f HgO + Br

(2)

ln

and/or

Hg + ClO f HgO + Cl

(3)

The most probable product of reaction 2 and/or reaction 3 is the formation of HgO, which has a very low vapor pressure (9 × 10-12 Pa (25)). Thus, the reaction would lead to the formation of TPM and not RGM as observed (6). Furthermore, in a thermodynamic study, calculations were carried out demonstrating that BrO and ClO with GEM are endothermic (26); thus, reactions 2 and 3 are most probably not important for the removal of GEM in the atmosphere. Instead of BrO and ClO, GEM may react with Cl and/or Br. Therefore, the data were analyzed assuming relative rate conditions between ozone and GEM. The method is widely used under laboratory conditions, where the reaction of interest is proceeding in competition with another reaction with a well-known reaction rate constant. So the system here is

O3 + X f product

(4)

Hg + X f product

(5)

and

where X is either Br or Cl and assuming that all other reactions are of negligible importance for the removal of ozone and GEM. The kinetic equations for reaction 4 and 5 are



[O3]t

[O3]0

dln[O3] ) -k4

∫ [X] dt t

0

(6)

and,



[GEM]t

[GEM]0

d ln[GEM] ) -k5

∫ [X] dt t

0

(7)

respectively. 2376

9

Integrating equation and dividing eqs 6 with 7, the relative rate expression is obtained.

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004

(

)

[GEM]0 [GEM]t

)

( )

[O3]0 k5 ln k4 [O3]t

(8)

A plot of ln([GEM]0/[GEM]t against ln([O3]0/[O3]t should give a straight line with intercept 0 and a slope equal to k5/k4. Figure 4 shows such a plot using the data from Station Nord. Data were selected for periods where the initial concentration of GEM was above 0.4 ng/m3 to ensure good signal-to-noise ratio and where three consecutive measurements of both ozone and GEM are decreasing. All measurements included were in periods with 24 h of daylight. There is a strong linear correlation (>99.9% confidence). A straight line is obtained by orthogonal regression with intercept close to zero as expected from eq 8. The slope is 1.44 and significantly different from 1 (>99.9% confidence). This is a very important result as a slope equal to 1 is expected, if the correlation was caused only by dilution due to mixing of an air mass containing, for example, background GEM concentrations with one depleted for GEM. Therefore, Figure 4 demonstrates for the first time that ozone and GEM have a mutual dependence that cannot be explained by meteorology but must involve chemical removal. The reaction rate constants of ozone with Br and Cl atoms in reaction 4 are very well-known due to their importance for stratospheric ozone depletion (27). Table 1 shows the calculated rate constants for GEM assuming a Cl reaction and Br reaction, respectively. In Table 1, the few available data from the literature of the reactions of GEM with halogen atoms are listed as well. All laboratory results are obtained using relative rate conditions. The reactions of halogen atoms with Hg are independent of temperature (28, 29); therefore, results of the various studies should be directly comparable. On the other hand, ozone reactions with halogen atoms are temperature dependent; thus, the rate constants obtained here of Hg are calculated at 233 and 263 K, which is representative for the conditions in Arctic during mercury depletion. In general, the half-life of GEM at Station Nord is 3-10 h during an AMDE. Using rate constants of the latest reaction

FIGURE 4. Natural logarithm to the relative concentrations of GEM and ozone during depletion episodes from 2000 to 2002. The regression analysis is carried out by orthogonal regression analysis. Only censored data are included where three consecutive measurements of both ozone and GEM are decreasing and where the initial concentration of GEM is larger than 0.4 ng/m3.

FIGURE 5. Comparisons between observed (black curve) GEM from 2000 to 2002 and calculated daily means of Hg0 for three model versions, one without depletion (orange) and two with depletion, where the blue curve is with a 3 h lifetime, and the red curve is with a 10 h lifetime during the depletion.

kinetic study (30), this lifetime corresponds to a concentration of Br or Cl at 1-3 and 0.1-0.3 pptv, respectively. Laboratories report Cl concentrations in the Arctic during AMDEs from 0.001 to 0.004 ppt (23, 31, 32), at least a factor of 25 lower in concentration than needed for the observed GEM depletion, whereas Br in the ppt level is reported by many authors (23, 24, 31, 32). This implies that Br most probably is the key species leading to mercury depletion.

On the basis of the results presented previously, the most important reactant for GEM removal is Br, and the reaction proceed with a second-order reaction rate constant of about 1 × 10-12 cm3 molecule-1 s-1. The previous result needs confirmation in the laboratory, and although the result is a strong indication of the removal channel for GEM, other possibilities need to be examined before a definite answer can be given. In particular, it is important to clarify the role VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2377

2378

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004

FIGURE 6. Comparison of measured boundary layer BrO column from the GOME satellite and the calculated concentrations of RGM for spring 2000. VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2379

FIGURE 7. Total annual average deposition of mercury without (left) Arctic mercury depletion and with (right) depletion in µg Hg/m2/year for the years 1999 and 2000. The total annual deposition north of the Polar Circle is indicated as well.

TABLE 1. Calculated Reaction Rate Constants for the Reactions between GEM and Br and Cl Based on Figure 4 and on the Reaction Rate Constants of Cl and Br with Ozone (kCl ) 1.7 × 10-11exp(-800 K/T) cm3 molecule-1 s-1 and kBr ) 2.9 × 10-11exp(-260 K/T) cm3 molecule-1 s-1 (27))a reactant X

kHg+X (10-12 cm3 molecule-1 s-1)

Kelvin

ref

Br Br Br Br Br Cl Cl Cl Cl

0.8 1.2 0.2 ×b1 0.08 0.3b 3.2 ×b1 0.4 14 16 15b 10 ( 4

233 263 295 120-170 298 233 263 120-170 298

this paper this paper 34 30 31 this paper this paper 29 31

a

a Rate constant at 233 and 263 K were used, representing the conditions in the Arctic. b Correct within a factor of 3.

of heterogeneous chemistry during AMDEs. Furthermore, there is a factor of 16 difference in the reaction rate constant of the reaction between Br and Hg0 reported (30, 33) and in both cases determined by the relative rate technique. The difference shows that secondary reactions in the laboratory systems clearly play a role; thus, the reaction rate constant needs to be determined by an absolute method. To the knowledge of the authors, there is not any study of the reaction rate constants of the reactions between Hg and ClO or BrO in the literature. There is good agreement between the theoretical calculated rate constant (26) and the rate constant extracted from the field measurements presented here. However, that might be a coincidence. The reaction between Br and GEM leads to the formation of a radical.

Hg + Br T HgBr• 2380

9

TABLE 2. Correlation between Measured and Calculated GEM for Each Year from 2000 to 2002a

(9)

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004

year

r2

2000 2001 2002

0.54 0.18 0.07

Model run is with a lifetime of GEM during depletion of 10 h.

The further fate of this radical is at the moment speculative, but the fate of mercury in the troposphere is the subject of the theoretical studies in refs 26 and 34. Furthermore, HgBr• may complicate the interpretation of the relative rate studies interfering with the assumption that only two reacts compete. Modeling. The DEHM model for mercury was run for October 1998 to October 2002. The model does a good job reproducing the occurrence and length of the period of AMDEs, especially for the year 2000, where the calculated and measured the period of AMDEs starts nearly at the same time and ends also at the same time. But the simple parametrization does not and cannot describe the fast variations during spring (Figure 5). In 2001, the results of GEM are quite similar except that there are larger concentration fluctuations during the AMDE period, which the models do not catch. For 2002, there is not a good agreement between the calculated and the measured GEM. Table 2 shows the correlation coefficients between the calculated and measured concentrations of GEM for each of the three different years. The highest correlation is found for 2000, less correlation in 2001, and no correlation in 2002. The main reason for the less agreement in 2001 and no correlation in 2002 between measured and calculated concentrations is probably that the assumption that AMDEs occur everywhere over sea ice with temperatures below -4 °C is too rough an assumption. In fact, the bromine has been connected to the formation and refreezing of leads (6). Therefore, the production of Br in the Arctic both varies geographically and in time that leads to a much larger fluctuation in the GEM removal

than predicted by DEHM. Therefore, the agreement is lower in 2001 and 2002 than in 2000 due to the larger GEM concentration fluctuation measured. A comparison of three model runs with observed GEM is shown in Figure 5. The three model runs are (1) without depletion, (2) with depletion where the lifetime during depletion is 10 h, and (3) with depletion where the lifetime is 3 h. The main difference between the two last runs with depletion is only a slightly higher minimum level of GEM in the case of a 10 h lifetime. The variations and duration of episodes are not changed. In the model, the removal of GEM leads to a build-up of RGM as observed in the field (6). The calculated concentrations of RGM have been compared with measured integrated surface column of BrO as obtained from the GOME satellite (35). The results from the GOME satellite are treated so that they are representative for the near surface concentrations (35). In Figure 6, the mean BrO column near the surface (in the boundary layer) and RGM concentrations for each month are shown for the period of January to June 2000. Figure 6 shows clearly that BrO and RGM have the same general temporal and geographical variability and reach their maximum level and extension in April to May. This finding supports that the conversion of GEM in fact is connected to sea ice with temperatures below -4 °C and to the chemistry of Br. Notwithstanding, there is generally good agreement between RGM and BrO, but some clear discrepancies can be observed. In May, the largest BrO concentrations are found along the coast of the Beaufort Sea (north of Canada and Alaska), whereas maximum RGM concentrations are predicted north of Greenland. At present, there is no explanation for this observation, but most probably it reflects the rough assumption that RGM is produced ubiquitously above surfaces with temperatures below -4 °C. Bromine is most probably formed on the surface of refreezing leads (6). These leads form and disappear again more or less randomly around the Arctic Ocean depending on the oceanic currents, wind, temperature, and solar flux. Therefore, large variations in the concentrations of bromine are expected during spring and thus also in the removal of GEM and the build-up of the RGM concentration. This feature is in fact clearly seen in the measurements of GEM (Figures 2 and 5), and it explains the discrepancy between the model results giving a smooth depletion event extending for the whole depletion period, whereas measurements show a long series of shorter depletion episodes during the depletion period. Unfortunately, we do not have similar figures of BrO for 2001 and 2002. They could have helped with an explanation of the larger difference between the calculated and the measured concentrations of GEM for these two years. The overall conclusion is that the main structure of AMDEs is reproduced, and the results present a large step forward in the understanding of the fate of mercury in the Arctic atmosphere. The reproduction of the main structure is a strong indication for the theory that the limiting factor is the surface conditions, which has to be sea ice with surface temperature below -4 °C. The average annual deposition of mercury for 1999 and 2000 is shown in Figure 7 for the Northern Hemisphere with and without AMDEs. The largest depositions are found close to the sources in Asia, Europe, and North America mainly due to the deposition of primarily emitted RGM and TPM that is removed fast mainly because of dry deposition and washout by rain. The calculations with and without AMDEs show again the importance of AMDE in the Arctic for the total deposition of mercury. Here, the photochemically formed RGM is removed mainly by dry deposition as the Arctic is characterized by its very dry climate. The total annual deposition increases in the whole Arctic, and for the area north of the Polar Circle, the total deposition of

mercury increases from 89 to 208 tons/year due to the depletion. While we believe that we have made a major advance in understanding the chemistry and meteorology governing AMDEs, these results associated with mercury load estimates have to be taken with caution for several reasons. (i) There is far from full understanding of the chemical processes controlling atmospheric mercury, and as a consequence, the parametrization of AMDE in the model is not at present adequate for a reliable quantitative calculation of the mercury burden. (ii) The source of Br is most probably from refreezing leads; thus, the temporal and geographical variability is not well-described by the model. (iii) Evidence is reported for reemission of mercury to the atmosphere, which is not included in the model. For these reasons, the 208 tons/year represents only an estimate of the upper limit for the mercury load to the Arctic area.

Acknowledgments A. Richter is acknowledged for providing us with the BrO data from the GOME satellite. The Danish Environmental Protection Agency financially supported this work with means from the MIKA/DANCEA funds for Environmental Support to the Arctic Region. The findings and conclusions presented here do not necessarily reflect the views of the Agency. The Royal Danish Air Force is acknowledged for providing free transport to Station Nord, and the staff at Station Nord is especially acknowledged for excellent support. Michael E. Goodsite was financially supported by NERI and the Danish Research Council.

Literature Cited (1) AMAP Greenland 1994-1996, Ministry of the Environment and Energy, Danish Environmental Protection Agency, 1997. (2) Grandjean, P.; Weihe, P.; White, R. F.; Debes, F. Environ. Res. Sec. A 1998, 77, 165-172. (3) Shotyk, W.; Goodsite, W.; Roos-Barraclough, M. E.; Frei, F.; Heinemeier, R.; Asmund, J.; Lohse, G.; Hansen, C. Geochim. Cosmochim. Acta 2003, 67(21), 3991-4011. (4) Lin, C.-J.; Pehkonen, S. O. Atmos. Environ. 1999, 333, 20672070. (5) Schroeder, W. H.; Anlauf, K. G.; Barrie, L. Y.; Lu, A.; Schneebeerger, D. R.; Berg, T. Nature 1998, 394, 331-332. (6) Lindberg, S. E.; Brooks, S. C.-J.; Lin, C.-J.; Scott, K. J.; Landis, M. S.; Stevens, R. K.; Goodsite, M.; Richter, A. Environ. Sci. Technol. 2002, 36, 1245-1256. (7) Berg, T.; Bartnicki, J.; Munthe, J.; Lattila, H.; Hrehoruk, J.; Mazur, A. Atmos. Environ. 2001, 35, 2569-2582. (8) Goodsite, M. E.; Brooks, S. B.; Lindberg, S. E.; Meyers, T. P.; Skov, H.; Larsen, M. R. B. Submitted for publication in Atmos. Environ. (9) Hansen, A. S.; Nielsen, T. G.; Levinsen, H.; Madsen, S. D.; Thingstad, T. F.; Hansen, B. W. Deep-Sea Res. Part I 2003, 50, 171-187. (10) Barrie, L. A.; Bottenheim, J. W.; Schnell, R. C.; Crutzen, P. J.; Rasmussen, R. A. Nature 1988, 334, 138-141. (11) Foster, K. L.; Patridge, R. A.; Bottenheim, J. W.; Shepson, P. B.; Finlayson-Pitts, B. J.; Spicer, C. W. Science 2001, 291, 471-474. (12) Christensen, J. Atmos. Environ. 1997, 31, 4169-4191. (13) Christensen, J. H.; Brandt, J.; Frohn, L. M.; Skov, H., Atmos. Chem. Phys., in press. (14) Heidam, N. Z.; Wåhlin, P.; Christensen, J. H. J. Atmos. Sci. 1999, 56, 261-278. (15) Impey, G. A.; Mikele, C. M.; Anlauf, K. G.; Barrie, L. A.; Hastie, D. R.; Shepson, P. B. J. Atmos. Chem. 1999, 34, 21-34. (16) Skov, H.; Egeløv, A. H.; Granby, K.; Nielsen, T. Atmos. Environ. 1997, 31(5), 685-691. (17) Ebinghaus, R.; Kock, H. H.; Schmolke, S. R. Fresenius’ J. Anal. Chem. 2001, 371, 806-815. (18) Skov, H.; Nielsdo´ttir, M. C.; Goodsite, M. E.; Christensen, J.; Skjøth, C. A.; Geernaert, G.; Hertel, O.; Olsen, J. Asian Chem. Lett. 2003, 7(2,3), 151-158. (19) Grell, G. A.; Dudhia J.; Stauffer D. R. A Description of the FifthGeneration Penn State/NCAR Mesoscale Model (MM5). NCAR/ TN-398+STR. NCAR Technical Note. June 1995; Mesoscale and VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2381

(20) (21) (22) (23) (24) (25) (26) (27)

(28)

Microscale Meteorology Division, National Center for Atmospheric Research: Boulder, Colorado, 1995; p 122. Petersen, G.; Munthe, J.; Pleijel, K.; Bloxam, R.; Vinod Kumar, A. Atmos. Environ. 1998, 32, 829-843. Pacyna E. G.; Pacyna J. M. Water, Air, Soil Pollut. 2002, 137, 149-165. Hausmann, M.; Platt, U. J. Geophys. Res. 1994, 99 (D12), 2539925413. Boudries, H.; Bottenheim, J. W. Geophys. Res. Lett. 2000, 27(4), 517-520. Tuckermann, M.; Ackermann, R.; Go¨lz, C.; Lorenzen-Schmidt, H.; Senne, T.; Stutz, J.; Trost, B.; Unold, W.; Platt, U. Tellus 1997, 49B, 533-555. Schroeder, W.; Munthe, J. Atmos. Environ. 1998, 32, 809-822. Goodsite, M. E.; Plane, J. M.; Skov, H. Environ. Sci. Technol. 2004, 38, 1772-1776. DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.; Kurylo, M. J.; Howard, C. J.; Ravinshankara, A. R.; Kolb, C. E.; Milina, M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric Modelling. Evaluation 12. NASA panel evaluation; 1997, p 266. Horne, D. G.; Gosavi, R.; Strausz, O. P. J. Chem. Phys. Lett. 1968, 48, 4758-4764.

2382

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004

(29) Grieg, G.; Gunning, H. E.; Strausz. J. Chem. Phys Lett. 1970, 52, 3684-3690. (30) Ariya, P. A.; Khalizov, A.; Gidas, A. J. Phys. Chem. 2002, 106, 7310-7320. (31) Ro¨ckmann, T.; Brenninkmeijer, C. A. M.; Crutzen, P. J.; Platt, U. J. Geophys. Res. 1999, 104, 1691-1697. (32) Jobson, B. T.; Niki, H.; Yokouchi, Y.; Bottenheim, F.; Hopper, F.; Leaitch, R. J. Geophys. Res. 1994, 99, 25355-25368. (33) Sommar, J.; Gårdfelt, K.; Feng, X.; Lindquist, O. Rate coefficient for gas-phase oxidation of elemental mercury by bromine and hydroxyl radicals. Paper presented at the 5th International Conference on Mercury as a Global Pollutant, Rio de Janeiro, 1999. (34) Calvert, J. G.; Lindberg, S. E. Atmos. Environ. 2003, 37, 44674481. (35) Richter, A.; Wittrock, F.; Eisinger, M.; Burrows, J. P. Geophys. Res. Lett. 1998, 25, 2683-2686.

Received for review June 18, 2003. Revised manuscript received December 15, 2003. Accepted February 2, 2004. ES030080H