Environ. Sci. Technol. 2003, 37, 5537-5544
Ambient Silver Concentration Anomaly in the Finnish Arctic Lower Atmosphere M. SHAMSUZZOHA BASUNIA AND SHELDON LANDSBERGER* Nuclear Engineering Teaching Lab, University of Texas, Pickle Research Campus, R-9000, Austin, Texas 78712 TARJA YLI-TUOMI AND PHILLIP K. HOPKE Department of Chemical Engineering, Clarkson University, Box 5705, Potsdam, New York 13699-5705 PAUL WISHINSKI Vermont Air Pollution Control Division, 103 South Main Street, Building 3 South, Waterbury, Vermont 05671-0402 JUSSI PAATERO AND YRJO ¨ VIISANEN Air Quality Division, Finnish Meteorological Institute, P.O. Box 503, FIN-00101 Helsinki, Finland
Mean silver concentrations in weekly particle samples collected at Kevo, northern Finland, were determined for the period of October 1964-March 1978 by neutron activation analysis. Two distinct periods were observed in the silver concentration levels over this time frame. During 19641970, mean weekly silver concentration levels were found in the range of 0.01-190 ng/m3 with an arithmetic mean of 2.19 ng/m3. A few very high silver concentration levels (>10 ng/m3) were observed in this period, some of which simultaneously occurred with some of the highest bromine and iodine concentration levels. During 1971-1978, silver concentration levels were in the range of 0.020.89 ng/m3 with a mean value of 0.09 ng/m3. The observed concentration levels in the later period matched well the data from the early 1990s reported at Sevettija¨ rvi, northern Finland, about 60 km east of Kevo. Data analysis, historical records for this region, and residence time analysis (RTA) using wind back-trajectories show that occasional smelting of silver-rich Norilsk ores at the Nikel smelter, Kola Peninsula, was probably a significant contributor to elevated mean silver concentration levels during 1964-1970. RTA alone was not able to unambiguously identify the most probable source region for highest silver impacts at Kevo due to the weekly integrated nature of the samples collected. Critical examination of wind backtrajectories (24 per day) for specific high silver, bromine, and iodine concentration weeks was carried out to supplement the ensemble RTA analysis (2 back-trajectories per day). The supplemental back-trajectory analysis revealed that deposition of the smelter component silver as well as the sea components (bromine and iodine) could occur together at Kevo during these weekly sampling periods. The study implies that data from weekly integrated samples are insufficiently time-resolved for RTA methods alone to * Corresponding author phone: (512)232-2467; fax: (512)471-4589; e-mail:
[email protected]. 10.1021/es034004q CCC: $25.00 Published on Web 11/11/2003
2003 American Chemical Society
unambiguously resolve the sources contributing to ambient atmospheric concentrations at Kevo, Finland.
Introduction Arctic air pollutants originate from various sources in Europe, Russia, far East Asia, and North America. Subsequently atmospheric transport of natural and anthropogenic emissions from mid-latitudes causes the Arctic haze phenomenon during the winter months in the High Arctic. Northern Finland, northern Norway, and northwest Russia inside the 66°32′ N latitude are known as the European sub-Arctic. This part of the sub-Arctic region has received attention in environmental research in the past decades mainly because of the large industrial activities beginning from the start of the last century. Some of the world’s largest emitters of heavy metals and sulfur dioxide are located in the Kola Peninsula, Russia. A wide variety of industrial activities, like ferrous and non-ferrous smelters or peat-fired power plants, have operated in this region for more than 60 yr. Heavy metal emissions are much higher for the non-ferrous mining, smelting, roasting, processing of nickel, copper, aluminum, etc. to the atmospheric environment than any ferrous industry (1). After World War II, the need for minerals and natural resources expedited the rapid expansion of these facilities under the socialist development policy until 1992 (1). Beginning in 1921, various ore bodies have been discovered in Kola Peninsula. The location of Kevo in northern Finland and parts of northern Norway and the Kola Peninsula are shown in Figure 1. In 1932, a nickel smelter was built at Nikel for smelting local ore bodies. In 1938, a larger nickel smelter was built at Monchegorsk. In 1965, a roasting plant at Zapoljarnij was built after discovering a good quality ore body in 1956 (2). Since 1938, at Monchegorsk good quality nickel ores of the local Pechenga area, near Zapoljarnij, were smelted. However, these ores were quickly exhausted, and very high quality ore bodies, rich with nickel, cobalt, silver, gold, and platinum, were discovered in Norilsk, northern Siberia. Norilsk is an isolated place and had no road or rail links to central Russia. The lack of a transportation system from Norilsk to major Russian cities and the extreme Siberian weather prohibited any attempt to develop a complete smelting facility at Norilsk. Thus, it was decided in 1964 to ship Norilsk ore to the Kola region for processing. Initially the transportation was seasonal. In 1969, ships from Norilsk reached the Kola Peninsula as late as November. Later the shipping season was lengthened from 5 to 10 months and resulted in an increase in ore processing from 600 000 ton in 1970 to 1 million ton in 1977 (3). Atmospheric emissions in the European sub-Arctic region resulted from the smelting of these different quality ore bodies in the Kola Peninsula smelters during the last century. While many episodic studies on particle chemical compositions have been reported for different locations around the circumpolar Arctic (4-8), there are a few data sets available with systematic long-term particulate matter collection and analysis (9-13). A long-term (1980-1995) systematic particle sampling and chemical characterization at Alert (82°5′ N, 62°3′ W) in the Canadian Arctic is reported in detail by several authors (14-16). Among these results, silver concentrations were only reported at Sevettija¨rvi, Finland; Ny A° lesund and Vardø, Norway; and North Earth Archipelago, Russia, from episodic studies (12, 13, 17). Historical long-term (October 1964-February 1978) weekly air particulate samples archived in Finland have been VOL. 37, NO. 24, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5537
TABLE 1. Location, Distance, and Industrial Activities around Kevo sampling location at Finland (a)
industrial locations (b)
approximate distance from a to b (km)
Kevo (69°45′ N, 27°03′ E)
Kirkenes, Norway (69°41′ N, 30°02′ E) Ga¨ llivare, Sweden (67°10′ N, 20°39′ E) Kiruna, Sweden (67°51′ N, 20°14′ E) Nikel, KPa (69°25′ N, 30°15′ E) Zapoljarnij, KP (69°30′ N, 30°43′ E) Murmansk, KP (69°02′ N, 33°15′ E) Kovdor, KP (67°35′ N, 30°40′ E) Monchegorsk, KP (67°51′ N, 32°48′ E) Olenegorsk, KP (68°08′ N, 33°30′ E) Apatity, KP (67°40′ N, 32°47′ E) Kandalaksha, KP (67°13′ N, 32°13′ E) Kirovsk, KP (67°36′ N, 34° E)
115
iron ore mine and mill
400
iron ore mine and mill
350
iron ore mine and mill
134
nickel smelter
145
nickel ore roasting
256 282
large harbor town with related industries, base of about 150 nuclear-powered vessels iron ore mine and mill
313
nickel, copper, and cobalt smelters
314
iron ore mining and mill
327
thermal power station, processing of apatite ore
352
aluminum smelter, nuclear power station
370
apatite open pit mine
a
type of industry
KP ) Kola Peninsula.
FIGURE 1. Locations of Kevo and important industrial cities in this region. analyzed for Ag, Al, As, Br, Ca, Cl, Co, Cu, I, In, K, Mn, Na, Sb, Se, Si, Sn, Ti, V, W, and Zn using neutron activation analysis (NAA) (18). These data provide the longest and most continuous set of concentrations ever reported for the subArctic region. Locations of the various types of industry and distances around the sampling location from Kevo are presented in Table 1. In this paper, only the silver concentrations at Kevo are presented and discussed relative to some other elements.
Sampling and Analysis The Finnish Meteorological Institute (FMI) collected total suspended aerosol particle samples in Kevo (69°45′ N, 27°02′ E, 98 m above the sea level) during the period of late October 1964 to early March 1978. Their primary purpose was to 5538
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 24, 2003
FIGURE 2. Schematic diagram of the aerosol sampling unit in Kevo. monitor for potential airborne radioactivity arising from Soviet nuclear tests. A schematic diagram of the sampling arrangement is presented in Figure 2. The sampling unit had two filters. Each filter was lead shielded and equipped with a Geiger Muller (GM) counter for radioactivity measurement. During a week-long sampling period, the airflows were alternatively directed for 4 h through each of the filters, and the aerosol radioactivity was continuously recorded from both online and offine filters. The air inlet was 7 m high above the ground. The particles were collected on Whatman 42 paper filters. The sampling procedure generated two filters per week, which were combined to give a total of 685 samples for the studied period (∼13.2 yr × 52 week). These samples were analyzed in a single batch to obtain the mean weekly concentration levels in the Kevo ambient atmosphere.
Silver in the filters was determined by epithermal NAA. Each of the filters was irradiated by epithermal neutrons for 1 min at the University of Texas 1-MW nuclear research reactor facility, followed by a 25-s decay time and 100-s counting time with a high-purity germanium (HPGe) γ-ray spectrometry system. Silver concentrations were determined by measuring the induced radioactivity of 110Ag (half-life ) 25 s) isotope in the aerosol filters. Irradiation and counting of all the samples were carried out within a 2-week period. The measurements were calibrated using a silver solution, and quality control was maintained by analyzing the standard reference material Gold Ore-CH2 from the Canada Center for Mineral and Energy Technology. Our result of 25.1 ( 2.6 ppm agreed very well with the certified value of 24.2 ( 2.0 ppm. Neutron flux was normalized for each of the sample irradiation days using sulfur powder irradiation. Other elements, like Cu, In, Sn, Zn, Br, I and Na, are determined in all the samples by thermal and epithermal NAA and discussed elsewhere (18).
Residence Time Analysis The Residence Time Analysis (RTA) technique is a probability metric to identify the potential source locations affecting the receptor site for a series of measured ambient species using a back-trajectory ensemble. The technique used in this work has been previously described (19, 20). Air parcel back-trajectories were computed using the HYSPLIT_4 model (21) developed at the National Oceanic and Atmospheric Administration (NOAA). The output of HYSPLIT_4 provided text files containing end points of hourly latitude, longitude, height, etc. of an air parcel. The use of these end points to develop a probability metric on a gridded array of cells in and around the receptor location is the key technique to locate the likely emission sources of air pollution. The metric first calculates a total residence time probability for each grid square considering all air parcel back-trajectories during the sampling period. Then a second residence time probability is calculated using only the back-trajectories associated with the high concentration subset sampling days for an element of concern (e.g., silver), called the high-value residence time probability. The difference between the highvalue and the total residence time probabilities represents the incremental probability for the concerned ambient element. Grid cells with the highest incremental probabilities signify potential source locations for the high concentration measurement days for that element. Finally, gridded array plots can be created for the incremental probabilities, producing a visual display to help identify the source region(s). The incremental probability can be calculated for a grid square (i,j) by the following equation: h TRIP% ) (TRp(i,j) - TRp(i,j)) × 100
where TR(i,j) is the total residence time for all the trajectories h over grid square (i,j); TR(i,j) is the total residence time for all the trajectories related to high subset samples over grid square (i,j) and can be calculated by l
TRp(i,j) ) (TR(i,j)/
m
∑∑TR
(i,j))
i)1 j)1
and l
h h TRp(i,j) ) (TR(i,j) /
m
∑∑TR
h (i,j))
i)1 j)1
In this work, a gridded domain of 500 × 500 cells with cell dimension 10 km × 10 km was used for RTA over and around
Kevo. Two back-trajectories were computed per day at 6:00 and 18:00 UTC at a starting height of 500 m using HYSPLIT_4 for a 5-day (120-h) period. Thus, for a weeklong sampling period, 14 back-trajectories represent the total air mass history for a particular sample. A total of 9608 backtrajectories were computed for the whole sampling period from 1964 to 1978 and used in RTA. The total residence time probability for each of the cells was calculated using all trajectories, while the high-value RT probabilities were evaluated for a subset of trajectories associated with the 7% highest silver concentrations for each of the 1964-1970 and 1971-1978 periods. The 2% highest silver concentrations were also examined during these periods in order to try to distinguish influence at the receptor on the very highest silver measurement days. Several high-subset cutoff values were considered, ranging from the highest 2% (very exclusive) to the highest 10% (inclusive of many moderate values) measurement weeks. For these data set of weekly integrated samples, the 7% level was the most inclusive cutoff used because it was the cutoff that began to resolve incremental probability patterns over the gridded region.
Results Measured Data. The measured weekly mean silver concentrations along with bromine and iodine in the Kevo atmosphere are presented in Figure 3 as time-series plots. As can be seen in Figure 3a, silver concentrations present two very different patterns during 1964-1970 and 1971-1978. Most of the silver concentrations were above the detection limit and had only 13% below detection limit (bdl) data points during 1964-1970. A few very high silver concentration levels (>10 ng/m3) were observed in this period, some of which simultaneously occurred with the highest bromine and iodine concentration levels in 1965, 1966, and 1969. However, in the period of 1971-1978, 73% bdl data points were found for silver concentrations at Kevo. In the earlier period, an arithmetic mean of 2.19 ng/m3 was found with a few high values around 100 ng/m3. In the later period, an average about 0.1 ng/m3 was observed with no value higher than 1 ng/m3. Virkkula et al. (17) reported average silver concentrations at Sevettija¨rvi, 60 km east from Kevo, of 0.09 and 0.13 ng/m3 for fine and coarse particles, respectively. Other silver data with the statistical results of the present work are presented in Table 2. As can be seen from Table 2, an average value of silver was 4.0 ( 10.0 ng/m3 in North Earth Archipelago, Russia (12). It was reported from a one-springmonth sampling period in 1988. It may be noted that this value was much higher than the other two measurements in 1985 and 1986 and possibly represented a special case. Statistically, this average value is comparable to 2.19 ( 13.92 ng/m3 for the period of 1964-1970 in the present work. A correlation study of 20 elements, determined in this work, showed a strong correlation of silver with bromine and iodine and a weak correlation with the other non-ferrous smelter elements such as copper, indium, tin, zinc, etc. It is observed that simultaneous occurrence of high silver, bromine, and iodine in some samples worked as leverage in the correlation study and showed a strong correlation among them, suppressing any significant correlation of silver with the other non-ferrous smelter elements such as copper, indium, tin, zinc, etc. The first 10 highest silver concentrations are presented in Table 3 with sampling periods along with the corresponding bromine and iodine concentration levels. It was found that measurements for sample numbers 233, 35, 232, 49, 87, and 230 represented some of the highest values of silver, bromine, and iodine in the database. The observations might be interpreted as implying that these three elements were coming from the same source and posed a great puzzle in this work. Also of interest is the fact that VOL. 37, NO. 24, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5539
FIGURE 3. Time-series plots of the silver, bromine, and iodine concentration levels measured in Kevo, northern Finland.
TABLE 2. Silver Statistical Data of This Work and Literature Data of the Arctic Region description this work, Kevo, Finland (69°75′ N, 27°02′ E) Sevettija¨ rvi, Finland (17) (69°35′ N, 28°50′ E) North Earth Archipelago, Russia (12) (79°5′ N, 95°4′ E) Ny A° lesund, Norway (13) (78°9′ N, 11°9′ E) Vardø, Norway (13) (70°4′ N, 31°1′ E) a
1964-1970 1971-1978 1992-1994 fine 1992-1994 coarse 1985 1986 1988 1983, 1984, & 1986 winters 1984 summer 1983 & 1984 winters 1984 summer
bdla
avg ( SD median avg ( SD median avg ( SD one sample avg ( SD avg ( SD avg ( SD median median median median
2.19 ( 13.92 0.31 0.09 ( 0.12 0.02 0.09 ( 0.05 0.13 0.12 ( 0.07 0.16 ( 0.26 4.0 ( 10.0 0.52 0.01 0.016
