Chemical composition of atmospheric aerosol in the European subarctic: Contribution of the Kola Peninsula smelter areas, central Europe, and the Arctic Ocean

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. D19, PAGES 23,681-23,696, OCTOBER 20, 1999

Chemical composition of atmospheric aerosol in the European subarctic: Contribution of the Kola Peninsula smelter areas,

central Europe, and the Arctic Ocean

Aki Virkkula, Minna Aurela, Risto Hillamo, Timo M•ikel•i, Tuomo Pakkanen, and Veli-Matti Kerminen

Finnish Meteorological Institute, Air Quality Research, Helsinki, Finland

Willy Maenhaut, Filip Fran9ois, and Jan Cafmeyer University of Gent, Institute for Nuclear Sciences, Gent, Belgium

Abstract. An 18-month set of concentration data of various elements in fine (diameter D < 2.5 gm) and coarse (2.5 gm< D <15 gm) particles is presented. Measurements were done at Sevettij'firvi, -60 km WNW of Nikel, a large pollution source on the Kola Peninsula, Russia. The concentrations in aerosol arriving from the Norwegian Sea and the Arctic Ocean are very close to the values observed at more remote Arctic sites. In air from the Kola Peninsula, approximately one third of the samples, concentrations of some trace elements were -2 orders of magnitude above the background concentrations. The elements most clearly transported in the pollution plumes from Kola Peninsula were Cd, Ni, Cu, V, Pb, As, Fe, and Co. Penner et al. [ 1993] presented a method for estimating black carbon (BC) emissions by comparing BC/SO2 (S) close to the sources and used a ratio 0.6 for the former USSR. We found that this ratio was < 0.1 in the

clearest pollution plumes from Kola peninsula. The ratio of the chemical mass to the gravimetric mass of the aerosol samples was -80% both for the fine and coarse particle filters, regardless of the source area. Comparison of the aerosol concentrations with the concentrations of elements in snow showed that the deposition was proportional to the aerosol exposure. The contribution of Kola Peninsula to the deposition is high, -80% for Ni, Cu, and Co and somewhat lower for other anthropogenic elements.

1. Introduction

Pacyna [1995] reviewed various studies on the sources, pathways, physical characteristics, and chemical composition of Arctic air pollution. The Arctic air pollution originates from various sources in Europe, Russia, and North America. On the European Arctic the Kola Peninsula industrial areas, Nikel, Zapolyarnyj, and Monchegorsk, are by far the largest source of sulphur and heavy metal aerosols [Tuovinen et al., 1993; Pacyna, 1995]. Maenhaut et al. [1989] investigated trace element composition and origin of fine particle (diameter D < 2.5 pm) atmospheric aerosol at two sites in the Norwegian Arctic, Spitsbergen, and Vfird0. The latter is located ~100 km NW from the pollution sources on the Kola Peninsula. Maenhaut et al. [1989] applied absolute principal component analyses (APCA) for source determination. APCA indicated for both sites that three

source components were contributing to the chemical composition of the aerosol. For Spitsbergen these were an anthropogenic, a soil dust, and a sea-salt component, and for V•d0 they were a mixed pollutant/soil dust component, a pollution component, which was identified as the Kola Peninsula, and a sea-salt component. The Norwegian Institute for Air Research (NILU) has conducted an extensive campaign on

Copyfight 1999 by the American Geophysical Union.

Paper number 1999JD900426. 0148-0227/99/1999JD900426509.00

measuring the dispersion of air pollutants from Nikel and Zapolyamyj [Sivertsen et al., 1992], and as a result of this campaign, Hagen et al. [ 1996] presented an exhaustive data set of heavy metal concentrations in atmospheric aerosol in the border areas of Norway and Russia in 1990-1995. Kelley et al. [1995] used a three-stage Berner impactor to measure aerosol size distributions of heavy metals on the Kola Peninsula in July and August 1993. Deposition of heavy metals around-the Kola has been investigated by Hagen et al. [1995], Jaffe et al. [1995], ,3iyri•s et al. [1996], Reimann et al. [1996], De Caritat et al. [1997], and Chekushin et al. [1998].

Atmospheric aerosols and trace gases have been measured at Sevettij'firvi (69ø35'N, 28ø50'E, 130 m above sea level) by the Finnish Meteorological Institute (FMI) since summer 1991. The original goal was to give information of air quality for the Lapland Forest Damage Project, a multidisciplinary effort to determine the effects of Kola emissions on the health of the

forests in Lapland [Tikkanen and Niemela, 1995]. Measurements performed at Sevettij'firvi have been presented by, e.g., DjupstrSrn-Fridell [ 1995], Kerminen et al. [ 1997, 1998], Virkkula et al. [1995, 1997), Virkkula [1997], and Maenhaut et al. [1999]. Virkkula et al. [1995] presented concentrations of major water soluble inorganic species measured at Sevettij•irvi in 1992-94. Parts of the same samples were later analyzed also for trace elements and in this work we present the results of these analyses.

The aim of the paper is to describe the chemical composition of atmospheric aerosol in the various air masses arriving at Sevettij•irvi: marine air, continental air, and in air polluted by the

23,681

23,682 VIRKKULA ET AL.: CHEMICAL COMPOSITION OF ATMOSPHERIC AEROSOL

Kola Peninsula industrial areas. We further give an estimate of the relative contributions of the various air masses to the samples. Chekushin et al. [1998] presented deposition data of various elements at many sites around the industrial areas of Kola Peninsula. One of their sampling sites was located close to our measurement site. We will investigate the relationship between concentrations of some elements in atmospheric aerosol and in the snow.

2. Instrumentation and Methods

Aerosol samples were collected using a virtual impactor (VI), which provides samples of two particle size ranges: fine and coarse, with aerodynamic diameter D < 2.5 gm and 2.5 < D < 15 gm, respectively. The upper limit was provided by the inlet. The total flow rate in the VI was 16.7 L min '• and the filters were

changed every 48 hrs. The total volume of air flown through a sample was 48 m 3. After being changed, the filters were stored at 4øC in Millipore Petri dishes.

The data consists of two sets. The first set consists of 165

samples taken between December 27, 1992, and December 1, 1993. The fine particles in the first set were sampled on Teflon filters (Millipore, •5 47 mm, pore size 3 gm) and coarse particles on polycarbonate membrane filters (Nuclepore, •5 47 ram, pore size 0.4 gm). On October 27, 1993 we started sampling with a second identical VI. The second set, altogether 100 samples, was finished on July 5, 1994. In the second sample set fine particles were also collected on polycarbonate membrane filters (Nuclepore, •5 47 mm, pore size 0.4 gm).

After sampling, filters were weighed using a Mettler M3 microbalance placed under a clean hood, cut into two pieces, and then weighed again to determine their percentages. At all stages filters were touched only with clean forceps and clean scissors, which were handled using powder free laboratory gloves. Filter pieces were analyzed for trace elements and for major inorganic ions. The mass of the blank polycarbonate filters varied so that 90% of them were between -20 and +20 gg and 50% of them were between -10 and +10 gg. For a 48-hour sampling with 16.7 L min '• this gives blank mass concentrations +0.42 (90%) and _+0.21 gg m '3 (50%). Relative humidity (RH) in the weighing room was neither controlled nor recorded for each filter, which decreases the accuracy of the mass determination. However, a 6-month daily recording of the humidity in the weighing room shows that the average RH _+ standard deviation was 40 _+ 10 % (minimum was 29% and maximum was 57 %).

For the whole sampling period, one part of the filters was analyzed for major inorganic species (8042-, NO3-, Cl-, NH4 +, Na +, K +, Ca 2+, Mg 2+) by ion chromatography, Dionex DX-100, at FMI. The other part was analyzed for various trace elements. All the fineparticle samples in the first set were analyzed at FMI using graphite furnace atomic absorption spectroscopy (GFAAS) for Cu, Cd, Ni, Pb, Cr, and V. Of the coarseparticle filters in the first set, only a selection of samples, altogether 38, were analyzed: those sampled during significant pollution episodes from Kola and some filters before and after these episodes. The second set of samples were analyzed at the University of Gent (UG) using three different techniques. The analysis of the trace elements was done using instrumental neutron activation analysis (INAA) and proton induced X-ray emission spectrometry (PIXE). The procedures are described in detail by Maenhaut and Zoller [1977], Maenhaut et al. [1981] and Maenhaut and Raemdonck [1984]. For each sample the INAA and PIXE results were combined into one data set. Because of element-dependent

differences in precision and in detection limits of the two techniques the PIXE data were selected for some elements, the INAA values were used for others, and the average of the PIXE and INAA values was used for other elements. This procedure is similar to that used by Maenhaut et al. [1993]. The INAA and PIXE analyses yielded concentration data of the following elements: Na, Mg, A1, Si, P, S, C1, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, As, Se, Br, Rb, Sr, Y, Zr, Nb, Mo, Ag, Cd, In, Sb, I, Cs, Ba, La, Ce, S m, Eu, Lu, W, Au, Pb, and Th. The

fine fraction filters sampled between April 2 and June 8, 1994 were completely pulverized and mixed up after the long INAA in adiation, which means that for this period we have no data for Sc, Co, As, Ag, Cd, Sb, Cs, La, Sm, Eu, W, and Au.

In addition, the filters were analyzed for black carbon (BC) using a light reflection technique, similar to that presented by Macias and Chu [1982] and used by Andreae [1983] and Andreae et al. [1984]. The instrument is a commercial smoke stain reflectometer (Diffusion Systems Limited, London, model 43). It was calibrated with the aid of over 100 filter samples for which the BC loading had been determined at Max-Planck- Institute (MPI) for Chemistry in Mainz, Germany, using an instrument and a procedure described by Andreae [1983] and Andreae et al. [1984]. The MPI reflectometer was calibrated using black carbon (soot) samples that were produced from burning acetylene, then filtering the soot and weighing the filters. The calibration curve of reflection absorbance (-log (reflection intensity of standard sample/reflection intensity of blank filter)) versus BC loading for the UG reflectometer exhibited excellent linearity, with a correlation coefficient of 0.984 [Maenhaut, 1997]. However, the BC concentrations obtained with the reflection technique differ from those obained with some other methods. For instance, when analyzing samples from biomass burning the BC concentrations appeared to be ~2-4 times higher than BC data obtained using a thermal evolution method, which lead to a conclusion that most likely the two methods measure different "black carbon" species [Maenhaut et al., 1996b].

Trace gas concentrations (SO2, NO2, and 03) were measured using a commercial differential optical absorption spectrometer (DOAS), AR500 manufactured by Opsis AB, Sweden. SO2 data is used in this study for an accurate timing of the Kola Peninsula pollution episodes. Performance and data of the DOAS measurements were presented earlier by Virkkula [1997] and Virkkula et al. [ 1997].

3. Source Area Determination

Transport routes were studied by calculating back-trajectories. Three or four three-dimensional 96-hour back-trajectories arriving at ground level, at 950 hPa, and at 900 hPa, were calculated for each day. They were calculated using the operational long-range Trajectory, Dispersion and Dose Model (TRADOS) of the Finnish Meteorological Institute [Valkama and Rossi, 1992; Valkama and Salonoja, 1993; Valkama et al., 1995]. The meteorological data applied by the model were from the numerical 6-hour weather forecasts of the Finnish version of the

Nordic High Resolution Limited Area Model (HIRLAM). The grid resolution of the HIRLAM fields is 55 km x 55 km. The trajectory set used in this study is the same as used by Virkkula et al. [1995, 1997].

The main source areas used in this study were the Norwegian and Arctic Seas (hereafter referred to as Sea), Kola Peninsula (hereafter Kola), and European Continent (hereafter Cont) (Figure 1). The filters were classified to these classes according

VIRKKULA ET AL.: CHEMICAL COMPOSITION OF ATMOSPHERIC AEROSOL 23,683

Figure 1. The source areas used for the filter classification: SEA, the Arctic Ocean and the Norwegian Sea; K, Kola Peninsula; CONT, continental Europe; NORDIC, Nordic Countries; NS, northern Siberia.

to the trajectories arriving at the site during the sampling and also according to SO2 concentration. The filter sampling time, 48 hours, is rather long, and therefore a sample often contains particles from various very different sources. For instance, when there is an anticyclone situated over Scandinavia, continental air from Europe and air from the British Isles flows clockwise over the North Sea and the Norwegian Sea, and, when arriving to the measurement site, it cannot be classified clearly as either marine or continental air. Therefore the class Cont contains only those trajectories that come from south of the Baltic Sea over either Scandinavia, the Baltic countries, or Russia and also the trajectories that cross the Baltic Sea. The last ones can be included in the Cont class since the sea-salt concentration in this

sea is low, and sea-salt particle concentration also decreases significantly because of the long transport route.

The class Kola contains all those samples that have either trajectories going over the Kola Peninsula or during which there was a clear sulfur dioxide peak. This class of samples is of diverse origin: In one of them, air has arrived at Kola from the Arctic Sea, and in the other it has arrived from the south without

going overseas during the last 96 hours. Furthermore, since this class contains all samples that are contaminated by Kola air, regardless of the length of the pollution episode, it has a wide range of concentrations from relatively clean samples (marine air plus short contamination) to very polluted samples (continental air arriving at Kola plus 48-hour pollution plume).

Contributions of the various source areas to the 161 filter

samples in the period from December 1992 to November 1993 were Sea, 53 (33 %); Kola, 56 (35%); and Cont, 13 (8%). In addition to these three main source classes, a finer classification showed that there were four (2.5%) samples that had their origin in the Nordic countries (NC) and five (3%) from northern Siberia over the sea (NS). The trajectories of the NC class came from the Norwegian Sea over Scandinavia and then to the site. The rest of the samples, 30 (19%), were of mixed origins. The contributions to the 100 filter samples in the period from October 1993 to July 1994 were Sea, 17 (17%); Kola, 35 (35%); Cont, 17 (17%); NC, 4 (4%); NS, 12 (12%); and mixed origins, 15 (15%). In both periods, there were two samples during which hourly averaged SO2 concentration was above 20 gg m -3 during the whole 48-hour

sampling period. These samples are regarded as the most representative of Kola air.

It was previously shown by Virkkula et al. [1997] that using hourly averaged SO2 concentration as criterion air at Sevettij•irvi is polluted by the Kola Peninsula sources -5-10% of time. Typical duration of these Kola events ranges from 2 to 24 hours. Even a short Kola event coinciding with our 48-hour filter sampling period makes a sample belong to the class Kola, explaining the rather high percentage of samples in this class.

4. Chemical Composition of Aerosol Particles

4.1. Comparison of Analytical Methods

The analysis of the filters with two methods provides a quality check of the results. For S, C1, and Na the concentrations are the

same within the accuracies of the methods (Figure 2). However, the sulfur analyzed using PIXE (SmxE) should be higher than the sulfur analyzed using IC as sulfate (S•c), because particles usually also contain other sulfur compounds like methanesulfonic acid (MSA'). Indeed, there is a difference in non-sea-salt sulfur (nss S, calculated as Smeasurea - 0.084 Nameasurea, using the S to Na ratio in the average sea water by Riley and Chester [1971]) concentrations. In the cleanair ([SO42'] < 1 gg m '3) fine particle samples, nss SmxE was, on the average, 9 and 21% larger than nss Sm in winter (November-February) and summer (May and June), respectively. For coarse particles the respective figures were-4 and 20%. There is a high uncertainty in the coarse particle nss S concentration difference, though, because the absolute concentrations were very low. The summer value is essentially the same as the MSA to nss SO4 ratio of 0.22 observed by Leck and Persson [1996] in the Arctic marine boundary layer in summer. Kerrninen et al. [1997] showed that the overall MSA to

nss SO42' ratio in aerosol at Sevettij•irvi ranged from 0.02 in polluted air to 0.34 in clean conditions in summer. These comparisons suggest that the difference nss Smx• minus nss S•c may be attributed to MSA-. However, MSA' was not analyzed in these samples, and we cannot be sure of it. Therefore in the mass balance calculations below we will use the Smx• as sulfate.

4.2. Observed Concentration Levels

To show the scope of the data, time series of cor/centrations of selected species in fine fraction are shown in Figure 3. In addition to the concentrations determined from the filters, the maximum

hourly averaged SO2 concentration measured during each filter sampling period is shown. The peak SO2 concentration is a good indicator of a pollution episode, as was explained above. The time series show that (1) concentrations of some trace elements range over three orders of magnitude, (2) typically there are one or two clean air samples between pollution episodes, and (3) the annual cycles are not as clear as in more remote Arctic sites.

Averages, medians, and ranges of concentrations in the whole data are presented in Table 1. The activation analysis (INAA) yields somewhat different detection limits for each sample. This leads to complications in presenting the data if there are many samples with high detection limits (DL) with the concentration < DL and some samples with very low concentrations but > DL. For calculating the median, maximum, and minimum concentrations we used for those concentrations that were < DL

the DL as the value, with a prefix '<', and sorted the concentrations in an absolute order. This leads to situations

where, for example for Sc, we have presented a median <0.0027 and minimum 0.0007.


Na

• 6,•,• k=l.01 •E "" Co=-10.2 /

FINE PARTICLES

C1

o'•o k=l.00

400 ,/ 06 ,•o' ' 80o •

INAA, PIXE (ng m '3)

Sk ß ß 2000 = 0.97 / Co -- -26 ./

r 2 = 0.95'.•:

1000 /• , øo 1 ooo 2000

1 ooo

500

Na , k = 1.03 "• Co:-.4.6

øo 5oo 1 ooo

COARSE PARTICLES

1500 Co=14 •

1000

5OO

00 500 1000 1500 INAA, PIXE (ng m a)

lOO

50

o

s

• ' 5b ' •6o

Figure 2. Comparison of ion chromatography (IC) analyses and the combined instrumental neutron activation analysis (INAA) and proton induced X-ray emission (PIXE) analyses.

The ranges and average concentrations of Ni and Cu were approximately the same in both years. However, for 1993 only some of the coarse particle samples were analyzed for trace metals. The selection was made so that samples of all the Kola pollution episodes were analyzed but only some samples of cleaner periods. Therefore the coarse particle concentrations of the trace metals of the 1993 data set are biased towards higher concentrations.

Table 2 presents the averages and medians of concentrations of the elements and ions arriving from the main source regions, Sea, Kola, and Cont. For the Kola samples we present two values. In both years there were two sampling periods during which the SO2 concentration was all the time over 20 gg m -3. These samples can be considered as the most representative of Kola Peninsula pollution, and they are therefore presented in a separate column. The other Kola column contains all samples that are contaminated by Kola air, regardless of the length of the pollution episode. Average sulfate concentration is higher in the Cont sample set than in the "All contaminated by Kola" sample set, although the highest nss sulfate concentration, 8.0 gg m '3, was measured on March 19, 1993, in a Kola pollution episode which lasted for the whole 48-hour sampling period. In air masses from Scandinavia, not presented in Table 2, all non- seasalt concentrations were higher than in the marine air but lower than in continental European air. For instance, the average fine-particle nss SO 4 was 1.8 + 1.3 lag m '3 (8 samples).

The observation that sulfur concentrations are not higher in the all contaminated by Kola sample set than in the Cont sample set can be explained by the fact that the smelters in Kola Peninsula emit sulphur into the atmosphere primarily as SO2, the contribution of particulate sulfate being <2% of the total sulfur [Tuovinen et al., 1993]. It was also shown by Tuovinen et al. [1993] that the particulate sulfur concentrations are essentially the same at various locations in northern Fennoscandia. For

instance, in 1989-1991 the average sulfate concentrations in summer and winter were 1.7-1.8 lag m '3 and 2.1-2.4 lag m '3, respectively, at four different measurement sites along the

Finnish-Russian-Norwegian border area ranging from Oulanka EMEP station, >300 km south of Sevettij•rvi, to Svanvik, only -10 km WNW of the smelter in Nikel [Tuovinen et al., 1993]. The highest average sulfate concentration was not at Svanvik, which is closest to the 15iggest smelter. On the other hand, SO2 concentrations showed a clear gradient toward the smelters [Tuovinen et al., 1993]. Maenhaut et al. [1999] collected size- fractionated aeorosol samples during an intensive field campaign at Sevettij'firvi. They observed that sulfur was not correlated with SO2 and the smelter-related trace metals Ni, Cu, As, and Se so confirming that the Kola peninsula smelters are not a major source of primary particulate sulfur, as the smelters are so close- by that SO2 has not had enough time to be oxidized in a significant fraction.

4.3. Comparison With Other Sites

As mentioned above, heavy metal concentrations in atmospheric aerosol were measured by NILU at many sites in the border areas of Norway and Russia in 1990-1995 [Hagen et al., 1996]. One of these sites is at Svanvik, -10 km WNW of the smelter in Nikel and 50 km ESE of Sevettij'firvi. The predominant wind direction in the area is between 180 ø and 270 ø so most of

the time both sites are not exposed to polluted air from the smelter but from the Norwegian Sea and from continental Europe. In fact, Hagen et al. [1996] stated that at the Norwegian side of the Norwegian-Russian border the concentrations of the analyzed elements were only a little higher or at the same level as the concentrations at Birkenes in southern Norway.

The sampling at Svanvik was done using a two-filter sampler, which separates the aerosol in fine (D<2.5 gm) and coarse (D>2.5 gm) particles, and with an inlet which provided a 10 gm upper limit for the coarse particles' diameter [Hagen et al., 1996]. The sampling time in the 1990-1995 period was 2 or 3 days. However, only the samples from the days with the highest SO2 concentrations were analyzed separately, while the rest of the samples were mixed and analyzed as one sample for each month

VIRKKULA ET AL.' CHEMICAL COMPOSITION OF ATMOSPHERIC AEROSOL 23,685

ion during 000

100

1

lO

o.1

O.Ol - ..

-o-S04

.-•-BC

lOO

o.1 0.01 .............. ' ' ' '

-o-Cu

ß -I- Ni

IO0

10

• 1

c•0.1 0.01

0.001 i i i i i i i i

-,-v

1 ß

0.001 - . ,

-.-Cd

lOO

•.• -*-Pb 1

0.1

0.01

27/12/92 27/02/93 30/04/93 29/06/93 29/08/93 30/10/93 17/1 2/93 01/03/94 30/04/94

Figure 3. Concentrations of selected species in fine fraction aerosol and the maximum hourly averaged SO2 concentration during each 48-hour sampling period.

[Hagen et al., 1996]. Hagen et al. [1996] present seasonal average concentrations for winter, October-March, and summer, April-September. Figure 4 shows seasonal average concentrations of some elements both at Sevettij•irvi and at Svanvik. When comparing the data of Sevettij'firvi and Svanvik, we have to keep in mind that all averaging times are not exactly the same. The comparisons that cover exactly the same time periods at both sites are summer 1993 for mass, V, Ni, Cu, Cd and Pb and winter 1993-1994 for mass, V, Ni, Cu, and Pb. The winter 1992-1993 data from Sevettij•irvi starts on December 27, for mass, V, Ni, Cu, Cd, and Pb, i.e., 3 months later than the Svanvik data. The winter 1993-1994 data from Sevettij'firvi starts on October 27, 1993, for Mn, Fe, Co, Zn and As, i.e., approximately 1 month later than the Svanvik data. The last period, summer 1994, covers April 1 to July 3 at Sevettij•irvi but April to September at Svanvik. It further has to be kept in mind that the size ranges of the sampled particles were not exactly the same.

The comparisons that cover exactly the same time periods at both sites show that for fine particles most of the concentrations are almost identical at the two sites, especially for the fine

particle mass. Instead, the coarse particle concentrations are consistently higher at Svanvik. This may be explained by the high deposition rate of large particles: by the time a pollution plume from Nikel reaches Sevettij'firvi, most of the large particles have been deposited on the ground. It may also be an indication that Svanvik is more exposed to coarse particles released from a lower level in the smelter area.

Figure 4 shows that average fine particle vanadium concentrations at Sevettij•irvi are higher than at Svanvik. However, especially in winter, the concentrations are not much higher than at a more remote site, Spitsbergen: Median V concentration of the winter campaigns 1983, 1984, and 1986 at Ny ]•lesund, Spitsbergen, was 0.54 ng m '3, and the median of the summer 1984 campaign was 0.022 ng m '3 [Maenhaut et al., 1989]. The median V concentration at Sevettij'firvi was 0.40, 0.20, 1.18, and 0.40 ng m '3 in winter 1992-1993, summer 1993, winter 1993-1994, and spring/summer 1994, respectively. In winter the concentrations are close to those observed at Spitsbergen, but in summer they are higher by a factor of 10.

There was a clear increase in Pb concentrations in autumn

1993 both at Sevettij•irvi and at Svanvik. The average (median)


Table 1. Concentrations of the Elements and Ions in All the Data

Fine Particles (D < 2.5 gm) Coarse Particles (2.5 gm < D < 15 gm) NDL (NT) AVE + s.d. Median (Max, Min) NDL (N•,) AVE + s.d. Median (Max, Min)

Mass* 257 4089 + 3081 3260 (18017, 373) 257 1581 + 1344 1169 BC 100 (100) 386 _+ 412 261 (2161, 2) 71 (100) 21 + 18 16 Na 99 (100) 141 _.+ 111 109 (644, <0.6) 100 (100) 184 + 212 107 Mg 62 (100) 39 + 17 41 (94, 16) 60 (100) 49 + 35 38 AI 99 (100) 9.7 + 9.8 6 (55, <0.2) 99 (100) 12 + 13 9.7 Si 85 (100) 28 _+ 23 15 (102, 3) 88 (100) 32 + 30 20 P 0 (100) - - (- -) 16 (100) 3.5 + 0.7 <2.5 S 99 (100) 626 + 475 481 (2400, <5) 92 (100) 33 + 24 26 nss S 99 (100) 614 + 477 474 (2395, <5) 88 (100) 18 + 16 12 C1 63 (100) 85 + 173 3 (955, <1.7) 89 (100) 307 + 389 111 K 100 (100) 23 + 18 17 (100, 1) 99 (100) 11 _+ 9 9 Ca 99 (100) 11 + 6 9 (30, <1.6) 99 (100) 17 + 12 13 Sc 27 (75) 0.0029 + 0.0017 <0.0027 (0.008, 0.0007) 66 (100) 0.0034 + 0.0026 <0.0023 Ti 49 (100) 0.86 + 0.46 <0.44 (2.2, <0.20) 61 (100) 1.1 + 0.86 0.50 V 97 (100) 2.0 + 2.9 0.8 (18, <0.01) 83 (100) 0.10 _.+ 0.16 0.05 V-93 * 134 (165) 1.1 + 2.5 0.29 (22, <0.07) 34 (38) 0.33 + 0.33 0.25 Cr 8 (100) 3.3 + 1.4 <2.0 (5.5, <0.78) 18 (100) 3.8 + 2.8 <1.9 Cr-93 * 135 (165) 0.16 + 0.15 0.09 (1.0, <0.04) - - - Mn 88 (100) 0.63 + 0.67 0.34 (3.8, <0.024) 92 (100) 0.38 + 0.41 0.25 Fe 98 (100) 13 _+ 13 9.3 (64, <0.7) 99 (100) 21 + 23 14 Co 51 (75) 0.11 _+ 0.14 0.037 (0.6, <0.012) 74 (100) 0.098 + 0.15 0.034 Ni 72 (100) 1.9 + 2.3 0.54 (11, <0.2) 78 (100) 2.0 + 3.2 0.53 Ni-93 * 139 (165) 0.8 + 2.2 0.19 (23, <0.07) 36 (38) 2.7 + 4.2 1.5 Cu 71 (100) 1.8 + 2.8 0.36 (15, 0.09) 69 (100) 1.3 + 2.2 0.24 Cu-93 * 141 (165) 1.5 + 4.3 0.30 (35, <0.05) 35 (38) 2.7 + 3.4 1.4 Zn 96 (100) 3.4 + 3.3 2.3 (16, <0.3) 84 (100) 0.41 + 0.37 0.26 Ga 25 (100) 0.12 + 0.21 <0.07 (1.1, <0.01) 16 (100) 0.24 + 0.51 <0.062 As 73 (75) 0.12 + 0.10 0.09 (0.51, 0.01) 74 (100) 0.047 + 0.051 0.034 Se 52 (100) 0.23 + 0.15 0.17 (0.83, 0.05) 4 (100) 0.096 + 0.062 <0.19 Br 98 (100) 2.3 + 1.0 2.1 (5.9, 0.7) 54 (100) 1.3 + 1.1 0.77 Sr 15 (100) 0.37 + 0.12 <0.28 (0.55, <0.09) 35 (100) 0.48 + 0.24 <0.26 Mo 23 (100) 0.15 + 0.05 <0.22 (0.78, 0.07) 36 (100) 0.24 + 0.30 <0.18 Ag 6 (75) 0.09 + 0.05 <0.10 (0.26, 0.03) 1 (100) 0.13 ' <0.09 Cd 9 (75) 0.40 + 0.19 <0.40 (0.86, <0.17) 1 (100) 0.21 <0.37 Cd-93 * 153 (165) 0.044 + 0.082 0.019 (0.77, <0.003) 37 (38) 0.015 + 0.010 0.012 In 46 (100) 0.0036 + 0.0031 <0.0027 (0.017, 0.001) 8 (100) 0.0006 + 0.0002 <0.002 Sb 64 (75) 0.056 + 0.075 0.024 (0.37, <0.003) 41 (100) 0.021 + 0.016 <0.011 I 100 (100) 0.66 + 0.27 0.67 (1.8, 0.10) 93 (100) 0.088 + 0.069 0.071 Cs 49 (75) 0.029 + 0.012 0.026 (0.064, 0.008) 42 (100) 0.018 + 0.008 <0.019 La 19 (75) 0.021 + 0.009 <0.019 (0.039, <0.0073) 30 (100) 0.022 + 0.012 0.020 Sm 29 (75) 0.0017 + 0.0008 0.0017 (0.004, 0.0008) 51 (100) 0.0020 + 0.0014 0.0016 Eu 10 (75) 0.018 + 0.009 <0.009 (0.043, <0.003) 10 (100) 0.012 + 0.002 <0.009 W 26 (75) 0.054 + 0.041 0.044 (0.20, 0.0086) 20 (100) 0.030 + 0.027 0.030 Au 27 (75) 0.00062 + 0.00031 <0.0005 (0.002, <0.00009) 33 (100) 0.02 + 0.12 <0.0005 Pb 87 (100) 2.1 + 2.5 0.99 (17, <0.33) 45 (100) 0.18 + 0.18 <0.14 Pb-93 * 164 (165) 0.61 + 0.88 0.30 (7.3, <0.01) 36 (38) 0.20 + 0.16 0.18 NO3* 203 (253) 87 + 41 71 (242, <15) 168 (253) 56 _+ 41 29 SO4' 253 (253) 1592 + 1503 999 (8215, 86) 202 (253) 108 + 71 72 nss SO4' 253 (253) 1555 + 1510 964 (8200, 54) 202 (253) 35 + 46 20 NH4* 252 (253) 202 + 227 115 (1429, <9) 122 (253) 25 + 50 8.8 Mg* 222 (253) 20 + 17 13 (94, <2) 229 (253) 29 + 29 16 Na* 250 (253) 150 + 138 107 . (788, <9) 236 (253) 243 + 248 126 CI* 147 (253) 227 + 257 69 (1571, <12) 200 (253) 459 + 499 231

(8054, 316) (87, <1)

(1037, 1.3) (182, 9) (76, <0.2)

(161, 2.5) (5, <1.o)

(111, 0.8) (111, <7)

(1815, O.9) (44, <2) (60, 0.8)

(0.013, 0.0005) (4, <0.36)

(1.2, 0.0005) (1.7, <0.02) (14, <0.69)

(3, <o. 15) (133, <0.55

(0.74, 0.009) (21, 0.04) (21, <0.07) (14, 0.06) (14, <0.05) (2.3, 0.02) (1.8, 0.006)

(0.26, 0.002) (0.75, 0.007)

(5.8, 0.08) (1.2, <0.12) (1.8, <0.08)

(<0.29, <0.03) (<0.81, <0.15) (0.068, <0.001) (0.011, <0.0005) (0.07, <0.003) (0.39, 0.0006) (0.05, 0.004) (0.06, 0.007)

(0.008, 0.0005) (0.045, <0.003) (0.13, 0.0042) (0.72, 0.0001)

(1.o, <O. lO) (0.8, <0.018)

(280, <15) (358, <29) (298, <1) (480, <9) (154, <2)

(1354, <9) (3199, <12)

Concentrations are in ng m '3. Arithmetic means (AVE) and standard deviations (s.d.) are calculated using only those samples with concentration above detection limit. Nm is the number of samples above detection limit, and Nr is the total number of samples analyzed.

*For mass and water soluble inorganic ions, analyzed by ion chromatography, the whole period from December 27, 1992, to July 3, 1994. *For Cu, Ni, Cd, Pb, Cr, and V concentrations in the period December 27, 1992 - December 1, 1993 are presented with denotation X-93.

Black carbon (BC) contains the statistics of all those samples where the given concentration was positive.

Pb concentrations in the fine fraction filters at Sevettij•irvi were 0.80 (0.36), 0.44 (0.26), and 1.77 (0.85) ng m -3 in winter 1992-1993, summer 93, and winter 1993-1994, respectively. However, Pb concentrations are low: For instance DjupstrOm et al. [1993] measured 4.0 ng m '3 in fine particles at Barrow, Alaska, in spring 1986. At Ny •lesund, median Pb concentration of winter campaigns in 1983, 1984 and 1986 was 3.0 ng m '3 [Maenhaut et al., 1989]. In summer 1991, median fine particle (D < 2 gm) Pb concentration measured over the Arctic Ocean was

0.059 ng m '3 [Maenhaut et al., 1996a], i.e., lower by a factor of 10 than that at Sevettij'firvi. The source of Pb is mainly in the exhaust gases {)f cars. The fact that the median Pb concentrations were higher at the more distant sites in the mid-1980s than at the more polluted site, Sevettij•irvi, 10 years later, is probably a result of the introduction of non-leaded fuel in the traffic at the end of the 1980s.

Ni, the element most evidently arriving from Nikel, rises clearly above the concentrations measured at more remote Arctic


Table 2. Average and Median Concentrations of the Elements and Ions Classified According to Source Areas All Contaminated

Sea Kola Only by Kola Cont NoL(Nr) AVE + s.d. Median NoL(Nr) AVE + s.d. No•(Nr) AVE + s.d. Median No•(Nr) AVE + s.d. Median

Fine Particles (D < 2.5 #m ) Mass 67 1943 + 1266 1585 4 12289 + 2900 87 5196 + 3193 4300

BC 17 (17) 71 + 66 42 2 (2) 1499 + 231 35 (35) 519 + 533 346 Na 17 (17) 212 + 163 153 2 (2) 49 + 15 35 (35) 126 + 91 93 AI 17 (17) 5.1 + 5.5 3.2 2 (2) 13 + 1 35 (35) 12 + 11 10 $i 14 (17) 14 + 15 7.0 2 (2) 52 + 9 33 (35) 37 + 26 28 $ 17 (17) 227 + 138 196 2 (2) 1674 + 34 35 (35) 788 + 460 643 nss $ 17 (17) 210 + 142 184 2 (2) 1669 + 32 35 (35) 778 + 462 639 CI 15 (17) 244 + 271 117 0 (2) <1.8 16 (35) 41 + 87 2.3 K 17 (17) 13 + 6 11 2 (2) 37 + 10 35 (35) 24 + 21 17 Ca 17 (17) 11 + 6 9.2 2 (2) 8.0 + 0.04 35 (35) 11 + 5 9.0 V 16 (17) 0.18 + 0.30 0.058 2 (2) 13 + 5 35 (35) 3.6 + 3.9 2.7 V-93 31 (53) 0.07 + 0.09 0.040 2 (2) 19 + 3 53 (56) 1.9 + 3.6 0.8 Cr 0 (17) - <1.9 0 (2) <2.3 2 (35) 2.3 + 0.5 <2.04 Cr-93 34 (53) 0.12 + 0.17 0.07 2 (2) 0.71 + 0.22 56 (56) 0.14 + 0.16 0.11 Mn 11 (17) 0.17 + 0.10 0.16 2 (2) 1.23 + 0.09 32 (35) 0.74 + 0.85 0.41 Fe 17 (17) 4.3 + 3.4 3.1 2 (2) 32 + 6 35 (35) 20 + 16 16 Co* 4 (12) 0.020 + 0.004 0.017 1 (2) 0.152 + 22 (24) 0.20 + 0.17 0.095 Ni 3 (17) 0.36 + 0.12 <0.18 2 (2) 5.7 + 1.5 34 (35) 3.2 + 2.7 2.1 Ni-93 35 (53) 0.063 + 0.062 0.02 2 (2) 16 + 7 56 (56) 1.5 + 3.2 0.7 Cu 4 (17) 0.21 + 0.09 <0.16 2 (2) 2.7 + 0.7 35 (35) 3.3 + 3.5 2.0 Cu-93 33 (53) 0.55 + 2.0 0.066 2 (2) 26 + 9 56 (56) 3.4 + 6.4 1.2 Zn 15 (17) 0.51 + 0.33 0.37 2 (2) 7.8 + 1.4 35 (35) 4.6 + 3.9 3.6 As* 11 (12) 0.055 + 0.026 0.040 1 (2) 0.25 + 24 (24) 0.19 + 0.14 0.14 $e 4 (17) 0.16 + 0.09 <0.11 2 (2) 0.43 ß 0.002 25 (35) 0.30 + 0.18 0.22 Br 17 (17) 2.20 + 0.88 2.06 2 (2) 2.0 + 0.2 34 (35) 2.0 + 0.9 1.7 Cd* 0 (12) - <0.4 1 (2) 0.27 9 (24) 0.40 + 0.19 0.34 Cd-93 41 (53) 0.008 ß 0.008 0.0057 2 (2) 0.55 + 0.22 56 (56) 0.090 + 0.120 0.064 $b* 8 (12) 0.021 ß 0.015 <0.012 1 (2) 0.081 23 (24) 0.092 + 0.11 0.042 I 17 (17) 0.66 + 0.20 0.65 2 (2) 0.74 + 0.45 35 (35) 0.69 + 0.35 0.67 Pb 9 (17) 0.32 + 0.15 <0.19 2 (2) 5.6 + 0.3 35 (35) 3.5 + 3.3 2.5 Pb-93 52 (53) 0.12 + 0.11 0.09 2 (2) 2.1 + 0.7 56 (56) 0.7 + 0.6 0.5 NO3 60 (67) 83 + 32 77 2 (2) 102 + 2 65 (87) 87 + 34 69 SOn 67 (67) 474 + 341 371 4 (4) 5832 + 1337 87 (87) 2240 + 1580 1859 nss SOn 67 (67) 427 + 340 314 4 (4) 5813 + 1333 87 (87) 2209 + 1582 1843 NI-h 67 (67) 73 + 41 57 4 (4) 436 + 248 87 (87) 216 + 198 152 Na 67 (67) 189 ß 164 129 4 (4) 79 + 49 85 (87) 122 + 102 84 CI 62 (67) 276 + 298 140 0 (4) <12 32 (87) 152 + 101 24

Coarse Particles (2.5 #m < D < 15 #m ) Mass 67 1847 + 1182 1540 4 1392 + 1114

BC 15 (17) 28 + 16 25 2 (2) 46 + 5 Na 17 (17) 391 + 270 351 2 (2) 17 + 9 AI 17 (17) 11 + 11 6 2 (2) 24 + 4 Si 16 (17) 24 ß 21 15 2 (2) 69 + 12 S 17 (17) 47 + 25 40 2 (2) 70 ß 41 nss-S 16 (17) 15 + 7 14 2 (2) 69 _+ 42 CI 17 (17) 661 + 504 575 1 (2) 6.7 K 17 (17) 18 + 10 17 2 (2) 6.6 + 2.7 Ca 17 (17) 22 + 12 23 2 (2) 22 + 1.84 V 9 (17) 0.057 ß 0.021 0.054 2 (2) 0.6 + 0.6 V-93 - - 2 (2) 1.5 + 0.2 Cr 4 (17) 2.7 + 0.9 <1.8 0 (2) <2.1 Mn 15 (17) 0.33 + 0.18 0.28 2 (2) 0.37 + 0.06 Fe 17 (17) 11.5 + 7.5 12.9 2 (2) 78 + 29 Co 8 (12) 0.031 + 0.013 0.033 2 (2) 0.25 + 0.18 Ni 8 (17) 1.36 + 1.44 <0.18 2 (2) 8 + 5 Ni-93 - - 2 (2) 18 + 2 Cu 9 (9) 0.27 + 0.11 0.16 2 (2) 5.0 ß 3.1 Cu-93 - - - 2 (2) 13 + 0.6 Zn 10 (17) 0.24 + 0.12 0.15 2 (2) 0.63 + 0.03 As 9 (12) 0.033 ß 0.009 0.036 2 (2) 0.076 ß 0.034 Se 1 (17) 0.15 + <0.29 0 (2) <0.31 + 0.06 Br 12 (17) 2.01 + 1.84 0.87 1 (2) 0.77 Cd 0 (12) <0.45 <0.53 0 (2) <0.37 Cd-93 - - 2 (2) 0.041 + 0.008 Sb 6 (12) 0.017 + 0.016 0.010 1 (2) 0.045 I 16 (17) 0.14 + 0.09 0.13 2 (2) 0.055 + 0.013 Pb 0 (17) <0.13 <0.13 2 (2) 0.47 + 0.12 Pb-93 - - - 2 (2) 0.35 + 0.02 NO3 31 (67) 40 + 28 13 3 (4) 32 + 10 SOn 67 (67) 116 + 67 87 4 (4) 121 + 65 nss 204 67 (67) 20 + 22 14 4 (4) 110 + 72 NH4 24 (67) 16 + 7 <9 3 (4) 27 + 10 Na 67 (67) 353 + 263 329 2 (4) 87 + 26 CI 65 (67) 630 + 569 536 2 (4) 26 + 5

28 6670 + 3501 5886

17 (17) 643 + 338 641 16 (17) 87 + 42 77 16 (17) 13 + 10 5.7 15 (17) 30 + 21 11 16 (17) 1007 + 519 907 16 (17) 1000 + 520 900 8 (17) 2.8 + 1.4 <1.8

17 (17) 34 + 19 34 16 (17) 12 + 6 11 16 (17) 2.7 + 1.8 1.5 12 (13) 1.9 + 1.8 1.3 3 (17) 4.4 + 1.5 <2.3

13 (13) 0.21 + 0.12 0.20 16 (17) 1.03 + 0.59 0.80 16 (17) 16.5 + 8.3 13 11 (14) 0.046 + 0.019 0.042 16 (17) 1.1 + 0.6 0.77 12 (13) 0.8 + 0.6 0.61 14 (17) 0.6 + 0.5 0.36 12 (13) 0.5 + 0.5 0.33 16 (17) 5.3 + 2.4 4.1 14 (14) 0.087 + 0.041 0.092 10 (17) 0.20 + 0.05 <0.20 16 (17) 2.53 + 0.87 2.3 0 (14) <0.35 + 0.18 <0.41

13 (13) 0.062 + 0.054 0.043 13 (14) 0.052 ß 0.041 0.035 17 (17) 0.61 + 0.25 0.62 16 (17) 1.9 + 0.9 2.0 13 (13) 2.0 + 2.0 1.3 19 (28) 83 + 44 42 28 (28) 2931 + 1718 2639 28 (28) 2914 ß 1718 2633 28 (28) 473 + 334 376 28 (28) 68 + 40 68

3 (28) 36 + 17 <12

87 1535 + 1299 1148

22 (35) 24 + 21 17 35 (35) 153 + 181 95 35 (35) 15 + 13 10 35 (35) 38 + 32 29 31 (35) 39 + 27 29 30 (35) 25 + 22 17 31 (35) 238 e 319 65 35 (35) 10.4 + 8.5 7.9 35 (35) 17 + 12 14 34 (35) 0.18 + 0.23 0.10 29 (29) 0.31 + 0.35 0.25 7 (35) 3.2 + 1.1 <1.9

34 (35) 0.43 + 0.43 0.31 35 (35) 34 + 32 26 34 (35) 0.18 + 0.19 0.09 35 (35) 3.4 + 4.1 2.5 29 (29) 3.2 + 4.5 1.7 34 (35) 2.3 ß 2.8 1.3 29 (29) 3.2 ß 3.5 1.9 32 (35) 0.46 + 0.30 0.38 31 (35) 0.079 ß 0.065 0.057 16 (17) 0.019 + 0.014 2 (35) 0.082 + 0.075 <0.23 0 (17) <0.20 + 0.07

18 (35) 1.07 ß 0.58 0.77 7 (i7) 0.86 + 0.50 1 (35) 0.21 <0.51 0 (17) <0.35

29 (29) 0.015 ß 0.011 0.012 5 (5) 0.011 ß 0.005 19 (35) 0.025 ß 0.019 0.012 9 (17) 0.018 ß 0.011 32 (35) 0.086 ß 0.070 0.068 15 (17) 0.046 ß 0.032 27 (35) 0.24 ß 0.20 0.18 8 (17) 0.12 ß 0.08 28 (29) 0.17 ß 0.10 0.14 5 (5) 0.44 ß 0.22 61 (87) 56 ß 34 44 22 (28) 72 ß 56 67 (87) 125 ß 76 98 28 (28) 67 ß 28 67 (87) 55 ß 63 40 28 (28) 41 ß 31 48 (87) 25 ß 42 16 16 (28) 26 ß 20 81 (87) 200 ß 209 101 20 (28) 61 ß 51 64 (87) 383 ß 422 228 9 (28) 115 ß 78

28 790 ß 615

9 (17) 23 ß 20 17 (17) 51 ß 46 16 (17) 11 ß11 14 (17) 27 ß 26 14 (17) 19 ß 13 13 (17) 15 ß 12 10 (17) 59 ß 55 16 (17) 7.1 ß 5.1 16 (17) 14 ß 12 14 (17) 0.064 ß 0.045 5 (5) 0.38 ß 0.08 3 (17) 6.57 ß 5.49

16 (17) 0.47 ß 0.64 16 (17) 18.9 ß 15.4 11 (17) 0.05 ß 0.04 13 (17) 1.2 ß 2.0 5 (5) 0.5 ß 0.6

12 (17) 0.39 ß 0.43 3 (5) 0.14 ß 0.12

15 (17) 0.43 ß 0.41

659

20

25

10

20

10

7.0

3.3

6.5

11

0.050

0.38

<2.1

0.25

12.2

0.03

0.3

0.2

0.18

0.08

0.26

0.015

<0.20

<0.71

<O.63

0.011

0.011

0.043

0.15

0.36

47

47

33

10

19

14

Concentrations are in ng m '3. AVE, s.d., Not., Nr, and elements presented with denotation X-93: see footnote of Table 1. *No fine fraction data for filters sampled between April 2 and June 8, 1994. See text for explanation.


[] Svanvik, fine [] Svanvlk, coarse ß Sevettijarvi, fine [] Sevettijarvi, coarse

(?, 4 E

=.2

0 I : I

2.0

1.5

1.0 0.5

0.0

2.0 Mn

(?, 1.5 E 1.0 • o.5

0.0

200 I Fe [• 150

400 50 0 • I '

0.3

0.2 • 0.1 0.0 • I •

c 3

0

8 T cu

4.0 T zn

m 3.0+ c 1.o

o.o • i t i i 0.8 m

I I I I ** c 0.2

0.0

0.08 T Cd

o.ot '- 0.02

0.00 i i i i

2.0 T Pb

1.0 c 0.5

0.0

Winter 1992/93 Summer 1993 Winter 1993/94 Summer 1994

Figure 4. Comparison of selected heavy metal concentrations measured at Sevettij•irvi (Sj) and Svanvik (Sv). The Sv data is taken from Hagen et al. [1996] and it covers the time from October 1992 to March 1993, April to September 1993, October 1993 to March 1994, and April to September 1994. One asterisk denotes Sj data from December 27, 1992, to March 31, 1993. Two asterisks denote Sj data from October 27, 1993, to March 31, 1994. Three asterisks denote Sj data from April 1 to July 3, 1994.

stations during air pollution episodes from Kola Peninsula. The highest Ni concentrations were >20 ng m 's, -100 times higher than in the background areas: median Ni concentrations measured by Maenhaut et al. [1989] in Ny •lesund were 0.29 ng m -s in winter and <0.2 ng m -s in summer. In summer 1991 the median Ni concentration over the Arctic Ocean was <0.08 ng m-3 [Maenhaut et al., 1996a]. At Sevettij•irvi the median Ni concentration was 0.19 ng m's in summer 1993 and 0.70 ng m's in winter 1993-1994.

The average and standard deviation of BC concentrations in the Sea, Kola, and Cont samples, 71 + 66, 519 + 533, and 643 +_ 339 ng m -3, respectively, fit well to the overall picture of BC concentrations in Arctic and continental air. For instance, at Ny •lesund, Spitsbergen, BC concentrations vary from 70 ng m '3 [Heintzenberg, 1982] to 293 ng m -3 [Clarke, 1989] in January and to 3 ng m '3 in July [Heintzenberg, 1982]. Typical rural BC concentrations are, e.g., 810 ng m -3 in Hungary [Heintzenberg and Meszaros, 1985] and 600 ng m '3 in Northern Michigan


[Cadle and Dasch, 1988]. Even though the data does not extend over even a whole year, it is clear that the BC concentrations have a clear seasonal cycle: the concentrations are higher in winter than in summer. The medians and ranges (minimum to maximum) of BC concentrations (ng m -3) in winter (November- February), spring (March-May), and summer (June) were 358 (91-2161), 273 (11-1355), and 46 (2-161), respectively. For further references, Penner et al. [ 1993] have compiled results of BC measurements at various locations in the whole world.

4.4. Elements Observed in the Pollution Plumes From Kola Peninsula

The data contains concentration time series of 46 elements and

compounds, including SO2. The strongest indication of a pollution plume from Kola Peninsula is high SO2 concentration. Figure 5 shows the scatter plot between SO2, averaged for the filter sampling periods, and the two elements with the highest correlation coefficient r in both data sets. In the first data set SO2 correlates best with Cd. The other elements that had r>0.6 with

SO2 were Ni, Cu, and V. In the second data set the elements that had r>0.6 with SO2 were Ni, V, Pb, As, Fe, Cu, and Co, in the order of highest to lowest correlation coefficient. For the second data set a correlation coefficient between Cd and SO2 cannot be calculated because of the high detection limit of Cd when using INAA: There were only nine samples above detection limit. However, all of these were measured in samples contaminated by Kola, compare Tables 1 and 2. In the first data set the correlation coefficient of Pb with SO2 is only 0.34, whereas in the second data set r = 0.78. This is in agreement with the observed increase of Pb concentration discussed above. It could be an indication of

increase of traffic in the Kola Peninsula, but it might also be explained by a change in fuel type or by some technical changes in the smelters. The correlation coefficient between SO2 and sea salt (Na) is slightly negative for both fine and coarse particles, showing that, usually, the air masses that arrive at Sevettij•irvi from Kola arrive at Kola from the continent, i.e., from the south.

Penner et al. [ 1993] presented a method for estimating global BC emissions by comparing measured BC/SO2 (S) ratios (SO2 presented as sulfur is SO2 (S)) in atmospheric aerosol and gas phase close to the emissions. The method is based on the fact that in most source-dominated environments ambient BC and SO2

concentrations are highly correlated. In urban locations the correlation coefficients vary between 0.81 and 0.98 [Penner et al., 1993]. For most areas Penner et al. [1993] used BC/SO2 (S) values between 0.2 and 1.1. Because of the lack of data for the

former USSR they used values measured in some eastern European cities to derive a ratio 0.6 for the entire former USSR. Practically all sulphur emissions north of the Arctic Circle originate from two well-defined regions: the Norilsk area in Siberia (2.2 Tg yr '• of SO2) and the Kola Peninsula (0.6 Tg yr -• of SO2). The sources in the Kola Peninsula contribute almost 20 % to the global anthropogenic sulphur emissions north of 60øN [Tuovinen et al., 1993]. Therefore for estimating the transport of BC to the Arctic using the BC/SO2 (S) ratio method it is essential to know this ratio close to the sources. Sevettij'firvi is -60 km from Nikel, and therefore the pollution plume concentrations can be regarded as representative of the BC/SO2 (S) ratios at the source. First, at Sevettij•irvi the BC and SO2 concentrations are not very highly correlated. For the samples in the Kola sector the correlation coefficient r was 0.52 when using a linear fit without an offset (Figure 6). The corresponding BC/SO2 (S) ratio was 0.059. When calculating directly from the actual ratios, the average (_+ standard deviation) BC/SO2 (S) ratio of the samples contaminated by Kola was 0.10 _+ 0.07. During the highest SO2 concentrations the ratio was close to 0.06. However, it is possible that this ratio is even lower: It was written above that for some

ambient air ,samples the light reflection technique has given higher BC values than some other methods. Therefore, when including the Kola Peninsula sources in a BC emission inventory, a BC/SO2 (S) value as high as 0.6 leads to overestimation of modeled BC concentrations in the Arctic atmosphere. The low correlation coefficient between SO2 and BC indicates that their sources are not necessarily the same. The Cu-Ni smelter in Nikel contains several electric smelters with 10 MW power consumption each, and the power consumption of the whole plant is estimated to be some hundreds of megawatts (R. Saarinen, personal communication, 1999). The authors of the present paper did not get not clear information of how this electricity is produced, but it is very probable that it is produced at the 1760 MW Kola Nuclear Power Plant at Polarnye Zori. It can be concluded that (1) the Kola emissions are "enriched" with SO2, with a BC/SO2 (S) ratio < 0.1, and (2) SO2 and BC do not have necessarily the same sources in the air arriving at Sevettij'firvi.

100 -

• 10

• 0.01

0.001

0.0001

0.1

December 27, 1992 - December 1, 1993 October 27, 1993 - July 3, 1994 100

Ni (ng m 4) = 0.19 SO2(pg m 'a) • V (ng m '3) = 0.24 SO2 (pg m 'a)

- i R2=0.73 •... • I I R• = 0'67 oCd t ß ß ß • ßNi o ß I I ß jm•'-" ß ,.,•. '"' ß _ ..•e•. '_',,' . ß •o• 8 o ø

ß o o oøø_ o

(y -

i•-. 0.1 o o

o 0.18 m 'a) ø o Cd(ng m 'a) = 0.00B5 SO2 (pg m '•) Ni (ng m 'a) = SO2 (pg O I R•1=0'75 I R• 0.01 I ? 0.70 I

I 10 100 0.1 I 10 100

SO•, (pg m '3) SO•,, pg m '3

Figure 5. Relationship of SO2 and the two elements that correlate best with SO2 in the two data sets.


10 --

0.1

BC = 0.059 SO2 (S)

R 2 = 0.27

'

0.01 I ß I I

0.1 1 10 100

SO2 (S), pg m '3

Figure 6. Relationship of black carbon (BC) and SO2 expressed as S at Sevettij•irvi in air contaminated by Kola Peninsula.

4.5. Enrichment Factors

Ratios of elements to A1 have been proposed as a method to determine the contribution of sources other than crustal material

in aerosol [e.g., Rahn, 1976; Hoff et al., 1983; Rahn, 1999]. Figure 7 presents crustal enrichment factors (EF^](X) = (X/A1)a•roso•t(X/A1)½rusta• rock) relative to A1 in average crustal rock [Mason, 1966], both in fine and in coarse filters. Comparison with the crustal enrichment factors for fine particles at Ny /•lesund and Vard0 [Maenhaut et al., 1989] shows that the enrichment factors follow the same pattern with some exceptions. DjupstrOrn et al. [ 1993] suggested that the elements with EF^i <

10 are nonenriched and that their natural cycles have not been perturbed by anthropogenic activities. In our fine particle samples the elements that meet this requirement are Sc, Si, Ti, Sm, Fe, Ca, La, and Mn, and in the coarse particle samples they are Si, Sc, Ti, Sm, Mn, Fe, K, La, Ca, and V. Of these elements, Fe in the fine and coarse particles and V in the coarse particles have high correlation coefficients with SO2. Thus, although according to the requirement above they should be classified to nonenriched elements, they clearly are transported from Kola also.

Similarily, a seawater enrichment factor was calculated from EFNa(J0 = (X/Na)aerosol/(X/Na)seawater, where (X/Na)seawater is the ratio of element X to Na in sea water [e.g., Hoff et al., 1983]. In fine particle samples the elements with EFNa(X)<I 0 were C1, Ca, Sr, Mg, Br, K, and in coarse particle samples they were C1, Br, K, Mg, Sr, S, Ca. However, the enrichment factor of C1 was <1. For the fine and coarse particles the average (+_ standard deviation) EFNa(C1) was 0.18 +_ 0.23 and 0.63 +_ 0.30, respectively. This is a common phenomenon observed at many sites; see Kerminen et al. [1998] for further references. Kerrninen et al. [1998] analyzed 10 Berner impactor samples at Sevettij'•vi and observed that the two things that influence chloride losses the most are the air transport time over continental areas and the concentration of sea-salt particles in air. In this larger data set these observations were confirmed. Figure 8 shows that chloride loss is the higher the lower the sea-salt concentration is. This is especially clear for the large particles. Both for fine and coarse particles, the chloride loss was highest in the continenal air. The fine particle EFNa(C1) was 0.45 +_ 0.25, 0.08 +_ 0.12, and 0.015 +_ 0.005 in the particles from the Sea, Kola, and Cont, respectively. The coarse particle EFNa(C1) was 0.92 +_ 0.17, 0.72 +_ 0.36, and 0.48 +_ 0.32, in the particles from the Sea, Kola, and Cont, respectively. C1 in the samples from northern Siberia was slightly more depleted than in the samples from the sea; EFNa(C1) was 0.86 +_ 0.12.

105

104

103

10'

10 ø

[.T.• 10"

Figure 7. Crustal enrichment factors (EF(X) = (X/A1)aeroso}/(X/A1)crustal rock) relative to A1 as compared to the ratios in the average crustal rock by Mason [ 1966]. The numbers below each element show the number of samples that have the concentration of both element X and A1 above detection limit.


1.1

1

0.9

0.8

• 0.7

• 0.6 z

u. 0.5

0.4

0.3

0.2

0.1

0

Fine particles Coarse particles

1.1

1.0

0.9

w ß 0.8

•0.7

ß ß '-• 0.6 z

ß ßß u. 0.5 ß ß w 0.4 ß ß

ß ßl ß ß ß ß % 0.3 0.2

ß

ß ß • .•,.•.._ 0.1 • --'- = , 0.0

1 O0 1000

Na, ng m '3

ß

ß ß ß,i' ; i 1

10 1 10 1 O0 1000

Na, ng m '3

0.01

0.001

0.1

0.01

ß

ß

II ß I I

ß ß ß ß

ß ß

[ i i i i

10 1 O0 1000 10000 1 10 1 O0 1000

nss S, ng m '3 nss S, ng m '3

Figure 8. Chloride depletion as a function of sea-salt (Na) and non-sea-salt sulfur concentration in fine and coarse particle samples.

In the fine particles, C1 depletion is probably due to the reaction of sea-salt particles with sulfuric acid: 2 NaC1 + H2SO4 • Na2SO4 + 2 HC1 as shown by the clear negative relationship between EFNa(C1) and nss S (Figure 8) and suggested, e.g., by Seinfeld and Pandis [ 1998]. In the coarse particles, C1 depletion is not related to nss S (Figure 8), and therefore the depletion should be explained by other reactions as discussed by Kerrninen et al. [1998]. They showed that the main constituents replacing chloride from supermicron sea-salt particles were sulfate and nitrate followed by MSA' and oxalate. However, in the present data set, there was, for some unexplained reason, no correlation between nitrate concentration and C1 depletion.

4.6. Mass Balance

For the mass balance calculation we grouped the analyzed elements and compounds in six groups: sea salt, nss sulfate, ammonium, nitrate, black carbon, crustal components, and other elements and compounds. Sea-salt concentration was calculated from Nameasured (1 +ZRi ss) = 3.256XNameasured, where Ri ss = Xi/Na is the ratio of concentration of element Xi to Na concentration in the average seawater by Riley and Chester [1971]. It was also calculated as Clmeasured + Nameasured(1 +•gi,no t Cl ss) = Clmeasured + 1.448 Nameasured, because C1 was depleted clearly compared to the average sea-salt composition as discussed above. The other elements were enriched compared to sea-salt ratios to Na, so the concentration of sea salt in the particles can be calculated from the equation shown above. Non-sea-salt sulfate (nss SOn) was calculated as 3 (S - 0.084 Na) from the sulfur measured by PIXE, although it was discussed above that the total sulfur also includes sulfur compounds other than sulfate, so the percentage of nss SO4 is somewhat overestimated, the maximum errors are in marine air

in summer, and the overestimation is probably of the order of 20%. The crustal mass was calculated from Almeasured (1 +•Ri crust) = 12.26 Almeasurea, where R/crust- Xi/A1 is the ratio of concentration of element Xi to A1 concentration in the average crustal composition presented by Mason [ 1966].

The time series of the mass balance (Figure 9) shows that, generally, the sum of the elements and compounds follows well the gravimetric mass, excluding some samples. The content of water in the samples is not known, since we did not weigh the samples in dry conditions and we did not record the humidity during weighing of each sample. The humidity in the weighing room is 40 _+ 10% so we may assume at least part of the missing mass is water. For instance, in the marine aerosol sampled by Mclnnes et al. [1996], water made up to 29% of the gravimetric mass at 47% relative humidity and 9% at 35% relative humidity.

Table 3 shows the mass balance of the filter samples presented as percentages of gravimetric mass of the filter, calculated for the whole data set and for the three major source areas, Sea, Kola, and Cont. For both the fine and coarse particle filters we have analyzed -80% of the gravimetric mass in all three air mass types. The missing mass may be water or organic matter, even in winter. A large part of this organic mass is probably associated with combustion aerosol. The source of BC is combustion and associated organic carbon (OC) concentrations are larger than BC concentrations in areas where fossil fuel sources are thought to dominate the emissions of BC concentrations [Liousse et al., 1996; Brernond et al., 1989; Cachier, 1992]. Liousse et al. [ 1996] assumed in their model that the ratio of organic matter to BC in fossil fuel combustion is 4.6. If we assume that the missing mass is organic carbon, the fine particle organic matter to BC ratio would be lower than that, -2


16

12

10

8

6

4

2

0 ,,

Fine particles • / Others

100 o•

80 •e I=;•INO3 (• =• Crustal

60 .• ISeasalt 40 • 20 • 1 BC

•NH4

E• nssS(SO4)

--=- Gravimetric mass

• Mass balance

.C, oarse particle,{s ,

7

6

3

2

I

,,

..... oo oooooooo

•oo• 8o •

•o .•

•o •

IB• Others

t NO3

1•3 Crustal

EEl Sea salt

/BC

B•NH4

E• nssS(SO4)

--=- Gravimetric mass

• Mass balance

Figure 9. Sum of the different chemical components, the gravimetric mass, and the mass balance in each filter. The mass balance is calculated from 100%(chemical mass/gravimetric mass).

in both Kola and Cont sample sets, and higher in the Sea sample set, and clearly higher for the coarse particles. As discussed below, part of the missing mass in the fine particles can also be biogenic secondary organic carbon. For instance, Li and Winchester [1989] showed that organic ions formate, acetate, propionate, and pyruvate altogether accounted for 20% of the total aerosol mass at Barrow, Alaska, in late winter.

It is obvious from Table 3 that the contribution of BC and

sulfate does not differ significantly in the aerosol that is transported from the continent from that contaminated by the Kola Peninsula. The contribution of sulfur is similar because the

Kola Peninsula smelters are not a source of primary particulate sulfur, as explained above (Section 4.2). It was also stated above that the slightly negative correlation coefficient between SO: and sea salt shows that, usually, the air masses that arrive at Sevettij'firvi from Kola arrive at Kola from the continent. However, the average contribution of sea salt is higher in the Kola sample set than in the Cont sample set. Therefore, in some of the Kola samples, air has gone from the ocean to Kola and then to the measurement site. This also explains the somewhat lower average contribution of BC in the sample set contaminated by Kola than in the aerosol from the continent.

The time series of the mass balance (Figure 9) shows that at the beginning of June 1994 the ratio of analyzed to gravimetric mass decreased systematically, both for fine and coarse particles. It was shown above that the BC concentrations decrease toward

summer. The contribution of BC to the total gravimetric mass also decreases. In the winter (November-February), spring (March-May) and summer (June) samples the medians and ranges (minimum to maximum) of BC contribution for the fine particle samples (in percent) were 11.4 (2.7-20.3), 7.2 (0.3-14.4), and 2.8 (0.1-5.3), respectively. Therefore the contribution of combustion-originated OC is likely to decrease as well. The

measurement station is located on a hill that is surrounded by a forest of birches and pines [Virkkula, 1997]. On May 30 the daily average temperature rose permanently above 5øC, which marks the start of the growing season in the Sevettij•irvi region. Pine

Table 3. Contribution of the Various Components to the Gravimetric Mass of the Filter Sample

All Samples, % Sea, % Kola, % Cont, %

AVE + s.d. AVE + s.d. AVE + s.d. AVE + s.d.

Fine Particles (D < 2.5 !•m )

aSS SO 4 43 + 13 32 + 16 49 + 7 50 _+ 5 NH4 5.1 + 2.9 3.9 + 2.3 5.0 + 2.6 7.3 + 2.6 Sea salt 17 + 16 35 + 22 12 + 11 6.1 + 4.7

(10 + 13) (26 + 21) (6.0 + 6.4) (2.7 + 2.1) Crustal 2.9 + 2.0 3.2 + 2.7 3.0 + 1.7 2.5 + 1.7

BC 8.3 + 4.8 3.5 + 2.8 9.3 + 5.1 12 + 3.2

NO 3 0.58 + 0.91 0.77 + 0.85 0.30 + 0.63 0.28 + 0.43 Others 1.3 + 0.6 0.91 + 0.47 1.5 + 0.6 1.2 + 0.4

Analyzed mass 78 + 17 79 + 20 80 + 11 79 + 8 (71 + 15) (70 + 19) (74 + 10) (76 + 7)

Coarse Particles (2.5 !•m < D < 15 !•m ) aSs SO 4 5.3 + 8.2 2.2 + 1.9 8.0 + 11 7.4 + 9.3 NH4 1.6 + 3.1 0.34 + 0.68 2.0 + 3.7 3.4 + 4.4 Sea salt 41 + 24 60 + 21 36 + 21 25 + 14

(34 + 24) (57 + 22) (29 + 22) (16 + 11) Crustal 18 + 24 8.4 + 9.3 24 + 33 24 + 17

BC 1.4 + 1.9 1.4 + 1.0 2.3 + 4.5 1.1 + 1.8

NO3 5.7 + 5.9 2.2 + 1.6 6.1 + 6.1 9.6 + 7.8 Others 3.9 + 4.9 1.3 + 0.9 6.4 + 6.9 4.6 + 3.2

Analyzed mass 77 + 42 76 + 15 85 + 58 75 + 32 (70 + 39) (73 + 15) (78 + 55) (65 + 29)

Kola contains all filters contaminated by Kola. The contribution of sea salt was calculated twice: by 3.256Nameasur• and Clmeasured + 1.448Nameasur•. The latter results are placed in parentheses.


Table 4. Aerosol Exposures During November 1, 1993 to March 31, 1994, and Deposition Velocities

E A, cms" B C

mg s m '3 NWS WS+NWS cm s" cm s" AI 225 0.7 1.1 0.35- 2.49 -

S 8619 0.013 0.39 0.02 - 0.17 0.3

SO2 (S) 38144 ....

SToT 46763 - 0.071 - - Ca 307 0.39 11 0.35 - 2.44 -

V 30.9 0.038 0.14 - 0.6

Fe 402 1.8 2.2 0.2- 1.44 1.3

Co 1.9 1.6 1.9 - -

Ni 40.7 2.1 2.7 0.01 - 0.03 0.5

Cu 28.3 1.3 2.0 - 1.3

As 1.6 2.3 3.2 - -

Pb 24.2 0.2 0.5 0.01 - 0.06 0.2

E is aerosol exposure defined in (1). A is the deposition velocity calculated in this work, B by Hillarno et al. [1993], and C by Foltescu et al. [1994]. NWS (non-water-soluble) is calculated by dividing the non-water-soluble fraction of the snow sample [Chekushin et al., 1998] by the exposure, and WS + NWS (water- soluble plus non-water-soluble) is calculated by dividing the total deposition [Chekushin et al., 1998] by the exposure. STow is the sum of the exposure of SO2 (S) and particulate sulfur.

trees emit monoterpenes, e.g., oc-pinene and I]-pinene [e.g., Janson, 1992], which, by oxidation, form condensable products that may either form new particles or condense on existing aerosol [e.g., Hatakeyarna et al., 1989]. Therefore we may assume that the trend in the mass balance of the fine particles is explained partially by biogenic secondary organic aerosol. On the other hand, this does not explain the decrease of mass balance for the coarse particles. Part of the decrease of the mass balance, for coarse particles probably all of it, may be explained by the increase of primary biological aerosol particles (PBAP): algae, spores of lichen, mosses, ferns and fungi (D >~1 pm) and pollen (D >~10 pm) [Matthias-Maser et al., 1996] after the melting of the snow cover in the region.

5. Aerosol Deposition

5.1. Estimation of Deposition Velocities

Chekushin et al. [1998] presented snow deposition in eight catchments located within the Barents region in Finland, Norway, and Russia in winter 1993-1994 and determined total yearly deposition rates for 16 elements. One of their catchment areas, Kirakka, was located only ~3 km away from our site, so we will use the deposition in snow at Kirakka to estimate deposition

' velocities of some elements. Dry deposition, Dd, is calculated by

10000

lOOO

lOO

lO

MW Ca: ..

V,

..... Dr -- 0.40 knu/a ,•, r:=0.99

1 10 1 O0 1000 10000

•, mg s m '•

FR

Fe.•

Ni, •'•a

V,•

SEi

D,• = 1.19 cm/s B, r a = 0.68

...... ...... i"o .... 'i"o .... i'o6"i œ, mg s m '•

10000

.• 1000

• 100

10

MW+FR •/ s •o:(•1 F

b.

D,• = 1.93 cm/s B, ra=0.92

œ, mg s m's

Figure 10. Comparison of aerosol exposure and deposition on snow. In the fits, only the data points have been used which are marked with solid squares. MW, meltwater; FR, filter residue.


multiplying the sampling time, concentration in air, and deposition velocity. For the whole sampling period Da can be thus estimated from

D d =•AtiCT, iVdep, i =Vdep•AtiCT, i =- VdepE (1) i--1 i--1

where Ati is the sampling time of sample i, Cr, i is total concentration which equals fine plus coarse particle concentration of sample i, vdep, i is the deposition velocity of sample i, n is the number of samples, vdcp is the average deposition velocity, and E is the aerosol exposure. E was calculated for a 5-month period, November 1993 to March 1994 (Table 4). The average monthly deposition presented by Chekushin et al. [ 1998] in the water-soluble fraction of the snow sample (meltwater (MW)), non-water-soluble fraction of the snow sample (filter residue (FR)), and their sum (TOT) of each element in snow were multiplied by 5. Table 4 shows the respective effective deposition velocities, calculated from (1), assuming that (1) only the non-water-soluble fraction and (2) both the water-soluble and the non-water-soluble fractions of the

elements found in snow got into it by dry deposition. Some of the deposition velocities are fairly high compared to other published •,alues. For a comparison we show in Table 4 deposition velocities at two other sites. Hillarno et al. [ 1993] determined the deposition velocities on the Greenland ice sheet at 0 and 99% relative humidity, and Foltescu et al. [1994] determined deposition velocities on a Teflon surface in southern Sweden.

Comparing these results, it has to be kept in mind that deposition velocity is not only highly dependent on the size of the particles but also on the nature of the deposition surface and meteorological conditions, like relative humidity and wind speed. In addition, the concentrations in snow are not only due to dry deposition of particles but also due to rainout and washout of the elements. The deposition velocities of Table 4 may thus be considered as upper limits.

For sulfur, there exists another large source of uncertainty. The average SO2 concentration in the period from November 1993 to March 1994 was 5.9 lag m -3, which gives an additional sulfur exposure of approximately 38 g s m -3. The sulfur deposition due to SO2 will be found in the meltwater, so an upper estimate of the deposition velocity of SO2 can be calculated assuming all the sulfur found by Chekushin et al. [1998] in the snow meltwater is due to SO2 dry deposition. We thus get an effective deposition velocity of 0.084 cm s -•. Tuovinen et al. [1993] used for SO2 a deposition velocity of 0.2 cm s '• in winter for modeling sulfur deposition. The comparison of the concentrations in snow and in the atmosphere suggests that this deposition velocity overestimates the sulfur deposition in snow.

5.2. Correlation of Aerosol Exposure and Concentrations Found in the Snow

$elin et al. [ 1991, 1992] and Foltescu et al. [ 1994] measured element concentrations in aerosol using a virtual impactor and dry deposition on a Teflon surface. They found nearly a linear relationship between the deposition velocity and the coarse/fine (C/F) ratio, when the cutoff for coarse particles is either 3.5 or 2.5 gm. In our case, the deposition surface is snow, so we cannot use their results as such. Nevertheless, we will use their

observation of a linear relationship between C/F ratio and deposition velocity by calculating a C/F-weighted exposure

C/F E=• •t•.' AtiCT, i (2) i=l

where Ati is the sampling time of sample i, Ci and Fi are the coarse and fine particle concentrations, respectively, and CT, i is the total concentration of sample i, and comparing that with the deposition presented by Chekushin et al. [ 1998]. Figure 10 shows the scatter plots of total (fine plus coarse) 5-month exposure versus 5-month deposition (water-soluble MW, non-water- soluble FR, and TOT, which is MW plus FR) on snow and the scatterplot of the C/F-weighted exposure versus total deposition. We fitted curves to the data points and used both a linear fit y = kx and a power curve y = kx • in all four cases. The results of the fits with the highest correlation coefficient r 2 are also shown in Figure 10. In all the fits we left out clear outliers (water-soluble Ca and non-water-soluble V), which would have made the fit very bad, and also sulfur, which is most probably not due to particle deposition only, as discussed above.

The plots show, in general, that the higher the exposure the higher the deposition. The water-soluble part (MW) of the deposition correlates better with the aerosol exposure than the nonsoluble part (FR). The total deposition also correlates well with the aerosol exposure. The best correlation of the total deposition in snow and aerosol concentrations is found between the C/F-weighted exposures and the total deposition. This is in agreement with the observation of Foltescu et al. [ 1994] that the deposition velocity can be roughly calculated from the C/F ratio multiplied by a constant. The high positive correlation between the aerosol exposure and deposition on snow suggests that both the water-soluble and the non-water-soluble fraction of the snow

samples can be explained by dry deposition of particles, although Chekushin et al. [1998] and Reimann et al. [1996] interpreted only the non-water-soluble part of the snow sample as deposited particles.

5.3. Contribution of Kola to the Deposition

It was shown above that the deposition is proportional to aerosol exposure, so the relative importance of the various sources to deposition can be estimated by calculating the exposure accumulated during sampling from the main source areas Kola and Cont and assuming that the deposition velocity is the same independent of the source. This probably underestimates the contribution of Kola Peninsula somewhat,

since nearby sources also contain large particles with higher deposition velocity.

Table 5 shows the results of such a calculation. For the first

sample period, December 27, 1992, to December 1, 1993, the exposures in Table 5 were calculated from fine particle data only because all coarse particle samples were not analyzed, but for the second sampling period, October 27, 1993, to July 3, 1994 the exposures were calculated from total concentration, fine plus coarse. Table 5 shows that the contribution of Kola is high, 70-80%, for Ni, Cu, and Co and somewhat lower for other elements.

6. Summary and Conclusions

An 18-month set of concentration data of various elements in

fine (D < 2.5 lam) and coarse (2.5 lam < D < 15 lam) particles at a subarctic site not far from Kola Peninsula pollution sources was presented. The element concentrations in aerosol arriving from the Norwegian Sea and the Arctic Ocean are very close to the values observed at more distant sites, e.g., Spitsbergen. In air from the Kola Peninsula, on the other hand, concentrations of

some trace elements were 2-3 orders of magnitude above the


Table 5. Exposures of Some Elements in the Two Data Sets and the Contribution of Kola and Continental Europe.

E, gg h m '3 Kola, % Cont, % December 1992 to November 1993

V 7.2 66 15

Ni 5.1 78 9

Cu 9.6 83 3

Cd 0.35 68 11

Pb 4.8 42 25

Cr 1.0 38 12

October 1993 to July 1994 V 10.5 63 22

Mn 4.7 41 26

Fe 172 54 16

Co 0.63 80 8

Ni 15 76 11

Cu 11 86 6

Zn 19.1 46 25

As 0.61 58 13

Pb 9.7 65 17

E is aerosol exposure defined in (1).

background concentrations. Approximately one third of the samples were contaminated by Kola Peninsula. The elements most clearly transported in the pollution plumes from Kola were Cd, Ni, Cu, V, Pb, As, Fe, and Co. The concentrations were

compared with the measurements run by the Norwegian Institute for Air Research (NILU) at Svanvik, -50 km from our site. For fine particles, most of the concentrations were almost identical at the two sites. In contrast, the coarse particle concentrations were consistently higher at Svanvik.

Penner et al. [1993] presented a method for estimating global BC emissions by comparing measured BC/SO 2 (S) ratios in atmospheric aerosol and gas phase close to the emissions. Because of the lack of data for the former USSR they used a ratio of 0.6 for the former USSR. This value leads most probably to overestimation of modelled BC concentrations in the Arctic

atmosphere because we found that the ratio was <0.1 in the clearest pollution plumes from Kola Peninsula.

The ratio of the sum of the chemical mass to the gravimetric mass was -80% both for fine and coarse particle filters, regardless of the source area. The missing mass may be water or organic matter because organic carbon was not analyzed. The decrease of the mass balance in the summer samples was a clear indication of increase of organic matter in the aerosol.

The atmospheric concentrations of some elements were compared with the concentrations found by Chekushin et al. [1998] in snow samples at a nearby sampling site. There was a high positive correlation between the aerosol exposure and deposition on snow. This suggests that both the water-soluble and the non-water-soluble fraction of the snow samples can be explained by dry deposition of particles. It was shown that the contribution of Kola Peninsula to deposition is high, 70-80%, for Ni, Cu, and Co and somewhat lower for other elements.

Acknowledgments. W.M., F.F., and J.C. are grateful to the Belgian Federal Office for Scientific, Technical and Cultural Affairs and the "Fonds voor Wetenschappelijk Onderzoek - Vlaanderen" for financial support. The project was also partially funded by the Academy of Finland (contract number 46906) and the Lapland Forest Damage Project.

References

Andreae, M.O., Soot carbon and excess fine potassium: Long-range transport of combustion derived aerosols, Science, 220, 1148-1151, 1983.

Andreae, M.O., T.W Andreae, R.J. Ferek, and H. Raemdonck, Long- range transport of soot carbon in the marine atmosphere, Sci. Total Environ, 36, 73-80, 1984.

Brtmond, M.P, H. Cachier, and P. Buat-Mtnard, Particulate carbon in the Paris region atmosphere, Environ. Technol. Lett., 10, 339-346, 1989.

Cachier, H., Biomass buming sources, in Encyclopedia of the Earth System Science, vol. 1, edited by W.A. Nierenmberg, , pp. 377-385, Academic, San Diego, Califi, 1992

Cadle, S.H., and J.M. Dasch, Wintertime concentrations and sinks of atmospheric particulate carbon at a rural location in northem Michigan, Atmos. Environ., 22, 1373-1381, 1988.

Chekushin, V.A., I.V. Bogytyrev, P. de Cadtat, H. Niskavaara, and C. Reimann, Annual atmospheric deposition of 16 elements in eight catchments of the central Barents region, Sci. Total Environ., 220, 95- 114, 1998.

Clarke, A.D., Aerosol light absorption by soot in remote environments. Aerosol Sci. Technol., 10, 161-171, 1989.

De Cadtat, P., C. Reimann, V. Chekushin, I. Bogatyrev, H. Niskavaara, and J. Braun, Mass balance between emission and deposition of airborne contaminants, Environ. Sci. Technol., 31, 2966-2972, 1997.

Djupstr6m-Fridell, M., Trace elements in airborne particles as observed at remote locations, Ph.D. thesis, Dept. of Phys., Chalmers Univ. of Technol., G6teborg, Sweden, 1995.

Djupstr6m, M., J.M. Pacyna, W. Maenhaut, J.W. Winchester, S.-M. Li, and G.E. Shaw, Contamination of Arctic air at three sites during a haze event in late winter 1986, Atmos. Environ., 27, 2999-3010, 1993.

Foltescu, V.L., J. Isakson, E. Selin, and M. Stikans, Measured fluxes of sulphur, chlorine and some anthropogenic metals to the Swedish west coast, Atmos. Environ., 28, 2639-2649, 1994.

Hagen, L.O., B. Sivertsen, and M. Johnsrod, Overvfikning av luft- og nedb0rkvalitet i grenseomrfidene I Norge og Russland. Oktober 1993- mars 1994, NILU Rep. OR 1/95, Norw. Inst. of Air Res., Kjeller, Norway, 1995.

Hagen, L.O., B. Sivertsen, and M. Johnsrod, Overvfikning av luft- og nedb0rkvalitet i grenseomrfidene I Norge og Russland. Tungmetaller i luft 1990-1995. NILU Rep. OR 28/96, Norw. Inst. of Air Res., Kjeller, Norway, 1996.

Hatakeyama, S., K. Izumi, T. Fukuyama, and H. Akimoto, Reactions of ozone with ot-pinene and [•-pinene in air: Yields of gaseous and particulate products, J. Geophys. Res., 94, 13,013-13,024, 1989.

Heintzenberg, J., Size-segregated measurements of particulate elemental carbon and aerosol light absorption at remote arctic locations, Atmos. Environ., 16, 2461-2469, 1982.

Heintzenberg, J., and A. Meszaros, Elemental carbon, sulfur and metals in aerosol samples at a Hungarian regional air pollution station, J. Hung. Meteorol. Serv., 89, 313-319, 1985.

Hillamo, R.E., V.-M. Kerminen, W. Maenhaut, J.-L. Jaffrezo, S. Balachandran, and C.I. Davidson, Size distributions of atmospheric trace elements at Dye 3, Greenland, I, Distribution characteristics and dry deposition velocities, Atmos. Environ., 27, 2787-2802, 1993.

Hoff, R.M., W.R. Leaitch, P. Pellin, and L.A. Barfie, Mass size distributions of chemical constituents of the winter Arctic aerosol, J. Geophys. Res., 88, 10,947-10,956, 1983.

Jaffe, D., B. Cerundolo, J. Rickers, R. Stolzberg, and A. Baklanov, Deposition of sulfate and heavy metals on the Kola Peninsula, Sci. Total Environ., 160/161, !27-134, 1995.

Janson, R., Monoterpenes from the boreal coniferous forest: Their role in atmospheric chemistry, Ph.D. thesis, Dept. of Meteorol., Univ. of Stockholm, Stockholm, 1992.

Kelley, J.A., D.A. Jaffe, and A.A. Baklanov, Heavy metals on the Kola Peninsula: Aerosol size distribution, Sci. Total Environ., 160/161, 135-138, 1995.

Kerminen, V.-M., M. Aurela, R.E. Hillamo, A. Virkkula, Formation of particulate MSA: deductions from size distribution measurements in the Finnish Arctic, Tellus, 49 Ser. B, 159-171, 1997.

Kerminen, V.-M., K. Teinil•i, R. Hillamo, and T. Pakkanen, Substitution of chloride in sea-salt particles by inorganic and organic anions, J. Aerosol Sci. 29, 929-942, 1998.

Leck, C., and C. Persson, Seasonal and short-term variability in dimethyl sulfide, sulfur dioxide and biogenic sulfur and sea salt aerosol particles in the arctic marine boundary layer during summer and autumn, Tellus, 48, Ser. B, 272-299, 1996.


Li, S.-M., and J.W. Winchester, Geochemistry of organic and inorganic ions of late winter Arctic aerosols, Atmos. Environ., 23, 2401-2415, 1989.

Liousse, C., J.E. Penner, C. Chuang, J.J. Walton, H. Eddieman, and H. Cachier, A global three-dimensional model study of carbonaceous aerosols, J. Geophys. Res., 101, 19,411-19,432, 1996.

Macias, E.S., and L.-C. Chu, in Particulate Carbon: Atmospheric Life Cycle, edited by G.T. Wolff and R.L. Klimisch, pp. 131-144, Plenum, New York, 1982.

Maenhaut, W., Composition and origin of the regional atmospheric aerosol at great distance from anthropogenic source areas: Assessment of the extent of the anthropogenic perturbation, in Belgian Impulse Prog. Global Change 1990-1996, final report, 105 pp., Fed. Off. for Sci., Tec., and Cult. Affairs, Brussels, 1997.

Maenhaut, W. and H. Raemdonck, Accurate calibration of a Si(Li) detector for PIXE analysis, Nucl. Instrum. Methods Phys. Res., Sect. B, 1, 123-136, 1984.

Maenhaut, W., and W.H. Zoller, Determination of the chemical composition of the South Pole aerosol by instrumental neutron activation, J. Radioanal. Chem., 37, 637-650, 1977.

Maenhaut, W., A. Selen, P. Van Espen, R. Van Grieken, and J.W. Winchester, PIXE analysis of aerosol samples collected over the Atlantic Ocean from a sailboat. Nucl. Instrum. Methods Phys. Res., 181,399-405, 1981.

Maenhaut, W., P. Cornille, J. Pacyna, and V. Vitols, Trace element composition and origin of the atmospheric aerosol in the Norwegian Arctic, Atmos. Environ., 23, 2551-2569, 1989.

Maenhaut, W., G. Ducastel, R.E. Hillamo, T.A. Pakkanen, and J.M. Pacyna, Atmospheric aerosol studies in southern Norway using size- fractionating sampling devices and nuclear analytical techniques, J. Radioanal. Chem., 167, 271-281, 1993.

Maenhaut, W., G. Ducastel, C. Leck, D.E. Nilsson, and J. Heintzenberg Multi-elemental composition and sources of the high Arctic atmospheric aerosol during summer and autumn, Tellus, Ser B, 43, 300-321, 1996a.

Maenhaut, W., I. Salma, J. Cafmeyer, H.J. Annegain, and M.O. Andreae, Regional atmospheric aerosol composition and sources in the eastern Transvaal, South Africa, and impact of biomass burning, J. Geophys. Res. 101, 23,631-23,650, 1996b.

Maenhaut. W., J.-L. Jaffrezo, R.E. Hillamo, T. M'akelii, and V.-M. Kerminen, Size- fractionated aerosol composition during an intensive 1997 summer field campaign in northern Finland, Nucl. Instrum. Methods Phys. Res., Sect. B, 150, 345-349, 1999.

Mason, B., Principles of Geochemistry, 3rd ed., John Wiley, New York, 1966.

Matthias-Maser, S., S. Gruber, and R. Jaenicke, The size distribution of primary biological aerosol particles in rain-water, in Nucleation and Atmospheric Aerosols, 1996: Proceedings of the Fourteenth International Conference on Nucleation and Atmospheric Aerosols, edited by M. Kulmala and P.E. Wagner, pp. 526-529, Pergamon, Tarrytown, N.Y., 1996.

McInnes, L.M., P.K. Quinn, D.S. Covert, and T.L. Anderson, Gravimetric analysis, ionic composition and associated water mass of the marine aerosol, Atmos. Environ., 30, 869-884, 1996.

Pacyna, J., The origin of Arctic air pollutants: Lessons learned and future research, Sci. Total Environ., 160/161, 39-53, 1995.

Penner, J.E., H. Eddieman, and T. Novakov, Towards the development of a global inventory for black carbon emissions, Atmos. Environ., 27, 1277-1295, 1993.

Reimann, C., H. Niskavaara, P. de Cadtat, T.E. Finne, M. •yriis, and V. Chekusin, Regional variation of snowpack chemistry in the vicinity of Nikel and Zapoljamij, Russia, northern Finland and Norway, Sci. Total Environ., 182, 147-158, 1996.

Rahn, K.A., Silicon and aluminium in atmospheric aerosols: Crust-air fractionation? Atmos. Environ., 10, 597-601, 1976.

Rahn, K.A., A graphical technique for determining major componenents in a mixed aerosol, I, Descriptive aspects. Atmos. Environ., 33, 1441- 1455, 1999.

Riley, J.P., and R. Chester, Introduction to Marine Chemistry, Academic, San Diego, Califi, 1971.

Selin, E., A. Mnubi, J. Isakson, V.L. Foltescu, and M. Djupstrtm, Transport and deposition of particulate pollution to the west coast of Sweden, Environment, Lifestyle, and Health FRN-Report 92: 3, pp. 56-71. Swedish Council for Planning and Co-ordination of Research, 1992.

Selin E., Djupstrtm M., Isakson J., and Mnubi A. (1991) A field study of the relationship between deposition rates and particle size, in Airborne Particles and Gases, and their Impact on the Cultural Heritage and Its Environment, Proceedings of the European Workshops, Ravello, December 1989 and 1990, edited by S. Hicks, U. Miller, S. Nilsson, and I. Vuorela, pp. 101-110, PACT, 33, 1991.

Sivertsen, B., T. Makarova, L.O. Hagen, and A.A. Baklanov, Air pollution in the border areas of Norway and Russia, NILU Rep. OR 8/92, Norw. Inst. of Air Res., Lillestr0m, Norway, 1992.

Tikkanen, E., and I. Niemel/i, Kola Peninsula pollutants and forest ecosystems in Lapland, final report, Lapland For. Damage Proj., Finland's Minist. of Agric. and For., Gummerms, Finland, 1995.

Tuovinen, J.P., T. Laurila, H. Lfittilii, A. Ryaboshapko, P. Brukhanov, and S. Korolev, Impact of the sulphur dioxide sources in the Kola Peninsula on air quality in northernmost Europe, Atmos. Environ., 27, 1379-1395, 1993.

Valkama, I., and J. Rossi, Description of the model TRADOS, in Evaluation of long range atmospheric transport models using environmental radioctivity data from the Chernobyl accident: The ATMES report, edited by W. Klug, G. Graziani, G. Grippa, D. Pierce, and C. Tassone, pp. 171-182, Elsevier, New York, 1992.

Valkama, I., and M. Salonoja, Operational long-range dispersion and dose model for radioactive releases, (in Finnish), Air Quality Reports, Finnish Meteorological Institute, Helsinki, 1993.

Valkama, I., M. Salonoja, H. Toivonen, J. Lahtinen, and R. P611'finen, Transport of radioactive gases and particles in the Chernobyl accident: Compadson of environmental measurements and dispersion calculations, in Proceedings of International Symposium on Environmental Impact of Radioactive Releases, Vienna, Austria, 8-12 May 1995, IAEA-SM-339, Int. At. Energy Agency, Vienna, 1995.

Virkkula, A., M. M'akinen, R.E. Hillamo, and A. Stohl, Atmospheric aerosol in the Finnish Arctic: Particle number concentrations, chemical characteristics, and source analysis, Water Air Soil Pollut., 85, 1997-2002, 1995.

Virkkula, A., R.E. Hillamo, V.-M. Kerminen, and A. Stohl, The influence of Kola Peninsula, continental European and marine sources on the number concentrations and scattering coefficients of the atmospheric aerosol in Finnish Lapland, Boreal Environ. Res., 2(4), 317-336, 1997.

Virkkula, A., Performance of a differential optical absorption spectrometer for surface 03 measurements in the Finnish Arctic, Atmos. Environ., 31,545-555, 1997.

•yriis, M., P. de Cadtat, V.A. Chekushin, H. Niskavaara, and C. Reimann, Ecogeochemical investigation, Kola Peninsula: Sulphur and trace element content in snow, Water Air Soil Pollut., 85, 749 - 754, 1996.

M. Aurela, R. Hillamo, V.-M. Kerminen, T. Makela, T. Pakkanen, and A. Virkkula, Finnish Meteorological Institute, Air Quality Research, S ahaaj ankatu 20E, FIN-00810 Helsinki, Finland. (e-mail: [email protected])

J. Cafmeyer, F. Francois, and W. Maenhaut, University of Gent, Institute for Nuclear Sciences, Proefiuinstraat 86, B-9000 Gent, Belgium. (e-mail: [email protected])

(Received January 29, 1999; revised May 20, 1999; accepted June 11, 1999.)

Chemical composition of atmospheric aerosol in the European subarctic: Contribution of the Kola Peninsula smelter areas, central Europe, and the Arctic Ocean

Documents