Four years of continuous total gaseous mercury - UQAM
Atmospheric Environment 36 (2002) 3735–3743
Four years of continuous total gaseous mercury (TGM) measurements at sites in Ontario, Canada P. Blancharda,*, F.A. Froudeb, J.B. Martina, H. Dryfhout-Clarkb, J.T. Woodsc b
a Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Ont., Canada M3H 5T4 Center for Atmospheric Research and Experiment (CARE), Meteorological Service of Canada, R.R. #1, Egbert, Ont., Canada L0L 1N0 c Envirotech Services, 196 Mason St., Bradford, Ont., Canada L3Z 1B2
Received 23 August 2001; accepted 13 May 2002
Abstract Total gaseous mercury (TGM) measurements have been conducted around the Great Lakes since 1997. At two of these sites (Egbert and Burnt Island), TGM concentrations presented significant seasonal variations. Possible explanations for the larger winter–spring/smaller summer–fall concentrations included seasonal meteorological differences, a northern hemispherical increase in coal combustion for wintertime heating, seasonal cycles of atmospheric oxidants and overall hemispherical source–sink relationship. The impact of populated/industrialized areas on the TGM concentrations at rural sites was demonstrated using pollution roses. Trend analyses for Egbert and Point Petre indicated relatively stable TGM concentrations between 1997 and 2000. Principal component analysis of TGM and trace metals confirmed the influence of the industrialized/populated area of Southwestern Ontario on both Egbert and Point Petre as well as metal recovery activities from Northern Ontario. For one episode of significant TGM concentrations, only a general geographical area could be implied as the source of atmospheric mercury. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: TGM; Atmospheric mercury; Trend; PCA; Great lakes
1. Introduction Mercury exists in the atmosphere predominantly in its elemental form (Hg0). Due to its stability, mercury has a residence time of the order of about 1 year and therefore can undergo long range transport (Schroeder and Munthe, 1998). Following deposition, mercury can be converted to methyl mercury, a highly toxic mercury species, which has been found to bio-accumulate in aquatic biota, especially fish. This has resulted in regulatory fish consumption advisories in various parts *Corresponding author. Tel.: +1-416-739-5701; fax: +1416-739-5708. E-mail address: [email protected] (P. Blanchard).
of North America and Europe, including the Great Lakes. Along with PCBs, mercury remains a contaminant of concern in fish in the Canadian waters of the Great Lakes (Scheider et al., 1998). To obtain spatial and temporal distribution of total gaseous mercury (TGM) levels on the Canadian side of the Great Lakes, a measurement campaign was started in 1997 at sites of the Integrated Atmospheric Deposition Network (IADN). The mercury measurements are part of the Canadian Atmospheric Mercury Network (CAMNet). The scope of the network and preliminary results from all sites have been described before (Kellerhals et al., 2000). In this paper we present the results of 4 years of continuous ambient TGM monitoring at two sites and 18 months at another site in Ontario, Canada.
1352-2310/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 2 - 2 3 1 0 ( 0 2 ) 0 0 3 4 4 - 8
3736
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
2. Experimental Total gaseous mercury was measured at Egbert (EGB) (44.22N, 79.78W), a rural site located about 65 km northwest of Toronto, since January 1997; Point Petre (PPT) (43.83N, 77.15W), IADN master station on Lake Ontario, since January 1997 and Burnt Island (BNT) (45.8N, 82.95W), IADN master station on Lake Huron since April 1998 (Fig. 1). At all stations, the measurements are made using Tekrant 2537A mercury vapour analyzers. This instrument has been described elsewhere (Schroeder et al., 1995; Poissant, 1997; Lee et al., 1998). In this paper the operational definition of total gaseous mercury (TGM) for Tekran-based measurements will be used (Schroeder et al., 1995). Briefly, after TGM was collected on a gold cartridge, it was thermally desorbed and detected by cold vapour atomic fluorescence spectroscopy. Two cartridges provided continuous measurements. The integration time was 30 min for 1997, decreased to 15 min in 1998, 1999 and 2000 at all stations. The instrument was calibrated daily using an internal mercury source verified quarterly by manual injections (Steffen and Schroeder, 1999). The data were quality controlled using RDMQt (Environment Canada, 1995) and were hourly averaged. Trace metals data have been collected at the sites as part of IADN since the early 1990s. Every 12 days, composite samples were collected for 24 h using a high volume sampler (PM 10) and subsequently analyzed
using inductively coupled plasma-mass spectrometry (IADN Technical summary, 1998). The available trace metals data for Egbert and Point Petre (1997, 1998) were used in conjunction with the TGM data for principal component analysis (Henry et al., 1984). This statistical method has been described before (Blanchard et al., 1997). Briefly, principal components were extracted from a correlation matrix and rotated by Varimax to facilitate interpretation. Components with eigenvalues greater than one were retained. This agreed with other criteria such as scree plot and analysis of variance. All computational calculations were conducted using SAS statistical software (version 6.12). For this analysis, TGM data were averaged to correspond to the daily sampling period of the trace metals measurements.
3. Results and discussion 3.1. Annual variability and co-pollutants Monthly TGM and temperature means are presented for the three sites in Figs. 2a–c for Egbert, Burnt Island and Point Petre, respectively. Similar seasonal variations are evident for the sites of Egbert and Burnt Island with a maximum in February–March and a minimum in August–September. (For both sites, air temperatures were significantly anti-correlated with TGM (R ¼ 0:21 for EGB and 0.26 for BNT).) The pattern of higher
N Burnt Island Ontario
Point Petre Lake H ron
Egbert Lake Ontario
Buoy Buffalo
Lake Erie Detroit
Fig. 1. Locations of the sampling sites.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743 4.0
15
2.5
10
2.0 5
1.5 1.0
0
0.5
-5
0.0 12/96 4.0
12/98
06/99
12/99
06/00
12/00 25 20 15
2.5
10
2.0 5
1.5 1.0
0
0.5
-5
0.0 12/96
-10 06/97
12/97
06/98
12/98
06/99
12/99
06/00
12/00 25
Point Petre
3.5
TGM, ng/m3
06/98
3.0
4.0
(c)
12/97
20
3.0
15
2.5
10
2.0 5
1.5 1.0
0
0.5
-5
0.0 12/96
Temperature, °C
TGM, ng/m3
(b)
-10 06/97
Burnt Island
3.5
Temperature, °C
20
3.0
Temperature °C
TGM, ng/m3
(a)
25
Egbert
3.5
3737
-10 06/97
12/97
06/98
12/98
06/99
12/99
06/00
12/00
Fig. 2. Monthly TGM and temperature means for (a) Egbert, (b) Burnt Island and (c) Point Petre. The boxes represent the 25th and 75th percentiles. The thick line is the mean and the fine line is the median. The error bars are the 10th and 90th percentiles.
Ozone, ppb 0
20
40
60
80
100
2.2
TGM, ng/m
3
2.0
1.8
1.6
1.4
PM2.5 Ozone SO2
1.2
1.0 0
10
20
30
40
PM2.5, SO2, µg/m
3
Fig. 3. TGM relationships with SO2, PM2.5 and Ozone for Egbert, 1997–2000.
3738
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
mercury concentrations in winter and decreasing in summer–fall has been observed before (Slemr and Scheel, 1998; Ames et al., 1998). Several factors might contribute to this behaviour, they include differences in meteorological conditions between summer and winter such as reduced mixing heights and higher wind speeds in winter, larger removal from the atmosphere by wet and dry deposition during warmer months. One other contributing factor might be the northern hemispherical wintertime increase in coal combustion for domestic heating purposes (Rotty, 1987). This is supported by the relationship seen at Egbert between TGM and daily sulfur dioxide (SO2) as shown in Fig. 3. Each point represents the average of 100 daily measurements. The agreement between TGM and SO2 (R2 ¼ 0:85 for a logarithmic regression) at Egbert likely results from similar sources, e.g. coal combustion for these two species. Lee et al. (1998) used continuous SO2 measurements as a tracer of local combustion sources in their examination of TGM concentrations in the UK. The seasonal cycle of atmospheric oxidant (e.g. ozone) might be also partly responsible for the seasonality of TGM at Egbert and Burnt Island although for ozone concentration X12 ppb (Fig. 3) there is no relationship between ozone and TGM. Segregation using water vapour mixing ratio (Poissant, 1997) was attempted but no significant correlation was obtained. However below 12 ppb, there is a negative relationship between TGM and ozone. This is likely due to pollution episodes originating from the industrialized/populated area of Toronto to the south–southeast of Egbert. Indeed nitric oxide data correlate well with TGM data (not shown), in effect titrating the ozone present in the anthropogenic plume reaching Egbert. The influence of the anthropogenic activities of Southern Ontario as sources of pollutants reaching Egbert is further established from the relationship between TGM and PM2.5 (Fig. 3) where, for this comparison, each point represents 1000 hourly measurements. In ‘‘clean air’’ situations (low levels of PM2.5) there is a lack of correlation due to background TGM concentrations. On the other hand, for PM2.5 concentrations larger than 15 mm/m3, other sources of PM2.5 unrelated to TGM gain importance, e.g. secondary formation of PM2.5. The result of this comparison is interesting since it implies that the correlation between TGM and PM2.5 could be used to indicate secondary production of aerosols. Finally the seasonal cycle at Egbert and Burnt Island might also be influenced by the behaviour of TGM on a hemispherical scale. Recent data of TGM vertical profiles (Banic et al., 2002) indicate that the Arctic springtime TGM depletion episodes could result in a significant hemispherical sink of TGM annually. The observations for BNT and EGB reflect this northern hemispheric behaviour of ‘‘background’’ TGM concentrations with a fairly strong seasonal cycle, albeit
influenced by local anthropogenic sources as discussed above. This is displayed in Figs. 4a–c where pollution roses are plotted for EGB, BNT and PPT respectively. Each bin represents 0.5 ng/m3 TGM, 301 wind direction with colours indicating frequency (%). Each bin has been normalized for overall wind direction frequency.
Fig. 4. TGM pollution roses for (a) Egbert, (b) Burnt Island and (c) Point Petre. Numbers in white are TGM concentrations in ng/m3.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
3739
At Egbert (Fig. 4a), episodes of larger TGM concentrations come from the industrialized/populated areas to the south–southeast of the site, as discussed above from the analysis of co-pollutants. The rose also shows that the lowest TGM concentrations are predominantly from the NNW which is consistent with clean air masses from Northern Canada. For Burnt Island (Fig. 4b), larger TGM concentrations are associated with air masses from the east to southeast again pointing to populated areas of Southern Ontario. TGM observations at Point Petre present a similar seasonal cycle (Fig. 2c) to Egbert and Burnt Island except that larger concentrations are superimposed to the annual cycles, particularly in summer 1998 and 1999. The pollution rose (Fig. 4c) points to the Southerly sectors as sources of TGM episodes for that site. Anthropogenic activities in New York state and Southwestern Ontario are likely the largest contributor to the TGM episodes observed at Point Petre but volatilization from Lake Ontario could also be of some importance (Poissant et al., 2000).
conducted for these data using a digital filter technique described by Nakazawa et al. (1997). Briefly, a long-term trend and seasonal cycle were determined by fitting a smoothing cubic spline and Fourier components to the data. The digital filter method has been used successfully to assess trends of greenhouse gases CO2 and CH4 (Worthy et al., 1998) as well as persistent organic pollutants such as PCB (Hung et al., 2001). Long and short term variations of the trend and the seasonal cycle were extracted using two cutoff periods: a short term cutoff period set to 4 months and a long term cutoff period set to 24 months. The results (Figs. 5a and b) indicate that for both sites, there is no clear trend for TGM between 1997 and 2000. In contrast, Slemr and Scheel (1998) found a decreasing atmospheric mercury trend between 1990 and 1996 for a site in Germany associated with large changes in regional emissions over that period. Although Canadian emissions decreased significantly between 1990 and 1995, current emissions have remained somewhat steady around 10 t/yr (Pilgrim et al., 2000).
3.2. Trend analyses
3.3. Source–receptor relationship
Four complete years of data were available for the sites of Egbert and Point Petre. Trend analyses were
Throughout the period, measurements of several trace metals have been conducted at the three sites as part of
4.0
Egbert
3.5
TGM, ng/m
3
3.0 2.5 2.0 1.5 1.0 0.5 0.0 1997
1998
(a)
1999
2000
2001
Year 4.0
Point Petre
TGM, ng/m
3
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 1997
(b)
1998
1999
2000
2001
Year
Fig. 5. Trends and seasonal cycles of TGM for (a) Egbert and (b) Point Petre: — trend, – seasonal cycle, + TGM.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
3740
trajectories for general anthropogenic/coal factor were similar. In the case of Egbert (Table 2), TGM is associated with two components (PC2 and PC4). Component 2 represent general anthropogenic activities and associated back-trajectories point again to the industrialized, populated areas of Southern Ontario. The other component containing a small loading of mercury (PC4) is associated with Co, Cr, Cu and Ni. Two events are responsible for this component and back-trajectories indicate air flowing from the northwest of the site passing over the Sudbury area where production of nickel and copper from sulphides ores mined in the area constitutes a significant source of atmospheric emission.
IADN. Results were available for 1997–1998 for Point Petre and Egbert. Principal component analysis was used to try and identify potential sources of metals to the two sites. This was done in conjunction with air parcel back-trajectories (Olson et al., 1978) incorporating geographical information to the statistical analysis. The results of the PCA analysis are presented in Tables 1 and 2 where loadings o0.3 have been omitted. The results of the analysis for Point Petre are similar to what has been obtained previously (Blanchard et al., 1997). The results include a crustal component (PC1), general anthropogenic/oil combustion (PC2), smelter (PC4) and another general anthropogenic/coal combustion component (PC3). The latter is particularly interesting since it has the higher loading for TGM associated with Ag, Cd, Cr, Mn, Mo, Se, Sn and Zn. For this component, three days Lagrangian back-trajectories associated with the highest 10% of scores were calculated corresponding to the mid-point of the sampling period (Fig. 6). These indicate air parcels passing over the heavily populated and industrialized areas of the lower Great Lakes to the southwest of the site. Factor analysis has been used previously on metals data combined with particulate mercury data (PM 2.5 and 10) from Perch River, NY in 1992–1994 (Ames et al., 1998). These authors found similar results to this analysis with industrial/coal combustion and smelter sources. Further, their back
Although the TGM concentrations varied only slightly over the years, some short term regional episodes of larger concentration deserve scrutiny. To discuss one particular episode, data from another site are needed. TGM levels from a buoy (43.24N, 79.26W) outfitted with a Tekran reached 70–80 ng/m3 on October 1–2, 2000 (Fig. 7). At Egbert (97 km north of the buoy), TGM concentrations reached 26 ng/m3 and were seen on the night of 2 October Point Petre (186 km northwest of
Table 1 Varimax rotated principal components analysis of 1997–1998 Point Petre trace elements and TGM data
Table 2 Varimax rotated principal components analysis of 1997–1998 Egbert trace elements and TGM data
PC1 Ag Al As Ba Cd Co Cr Cu Fe Mn Mo Ni Pb Sb Se Sr Sn Ti V Zn Hg Eigenvalues %Variance explained
PC2
PC3
PC4
0.48 0.92 0.65
0.36
0.31
0.41
0.65 0.87
0.88 0.78 0.41
0.32 0.39 0.85
0.41 0.69 0.92 0.82
0.33 0.83
0.79 0.50
0.76 0.35 0.34 8.5 40.6
0.81 0.38 (0.14) 3.2 15.4
0.35
0.67 0.73 1.7 8.0
0.54
1.5 7.0
PC1
PC5
0.90
0.66 0.84
3.4. Episode of significant TGM concentration
1.1 5.3
Ag Al As Ba Cd Co Cr Cu Fe Mn Mo Ni Pb Sb Se Sr Sn Ti V Zn Hg Eigenvalues % Variance explained
PC2
PC3
PC4
0.72 0.97 0.73 0.72
0.56 0.33
0.64
0.67 0.30 0.45
0.50 0.71 0.69
0.96 0.94 0.85 0.58 0.78 0.67
0.34 0.68 0.32
0.86
0.95 0.76 0.97 0.46 9.3 44.2
0.60 0.74 0.66 3.3 15.5
0.33 2.1 10.1
(0.25) 1.3 6.0
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
3741
Ontario
Point Petre
Fig. 6. Back-trajectories (950 mb) at Point Petre for the 10% of samples with the highest scores for PC3. Circles: 21 September 1998; squares: 17 June 1998; triangles: 28 July 1997 and crosses: 16 August 1998.
80 Egbert Point Petre Buoy
70
TGM, ng/m
3
60 50 40 30 20 10 0 0:00:00 Sept 30
0:00:00
0:00:00
0:00:00
Oct. 1
Oct. 2
Oct. 3
0:00:00
Fig. 7. TGM concentrations at the Lake Ontario buoy, Egbert and Point Petre during the October 2000 episode.
the buoy) experienced an increase in concentration simultaneously to the buoy, although the concentrations observed were much lower (peaking around 6 ng/m3). Wind direction was from the south with averaged speed 4–5 m/s. Concentrations for other gases such as CO, NO2, O3 and even PM2.5 remain relatively constant
throughout the period (J. Brook, personal communication). Back trajectories indicate that the air mass traveled over the Niagara area, eastern tip of Lake Erie, Buffalo area and Western Pennsylvania. The discrete nature of the TGM peak points to localized sources. Considering the industrialized nature of the Niagara
3742
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
region (Crittenden, 1997), it is foreseeable that the high levels of TGM observed at the buoy originated from that area.
4. Conclusion For two sites in Ontario, the seasonal pattern of total gaseous mercury was indicative of background northern hemispherical conditions superimposed with regionally influenced episodes. The examination of co-pollutants such as SO2, O3 and PM2.5 helped to assess whether seasonal patterns were due to factors such as winter coal combustion, seasonality of atmospheric oxidant and northern hemispheric source–sink relationship. No clear trend was apparent for the years of measurements in this study (1997–2000). Principal component analysis of TGM and trace metals data pointed to populated/ industrialized areas of Southern Ontario and metals recovery operations of Northern Ontario as sources of TGM episodes. The advantage of the short time resolution afforded by the continuous TGM analyzer used in this study was displayed during a significant episode of TGM concentration.
Acknowledgements The authors would like to thank the site operators D. Smith, F. Orford and particularly C. Green. C. Audette for the IADN trace metals data and D. Worthy for the use of the digital filter model. Dr. J.R. Brook for copollutants data from Toronto. Drs. K Puckett and C. Banic for helpful discussions.
References Ames, M., Gullu, G., Olmez, I., 1998. Atmospheric mercury in the vapor phase and in fine and coarse particulate matter at Perch River, New York. Atmospheric Environment 32 (5), 865–872. Banic, C.M., Beauchamp, S.T., Tordon, R.J., Schroeder, W.H., Steffen, A., Anlauf, K.A., Wong, H.K.T., 2002. Vertical distribution of gaseous elemental mercury in Canada. Journal of Geophysical Research, submitted for publication. Blanchard, P., Hopper, J.F., Hoff, R.M., 1997. Principal component analysis of IADN data from a site in Ontario. 97-A805, Presented at the Air & Waste Management Association’s 90th Annual Meeting & Exhibition, June 8– 13, 1997, Toronto, Ont., Canada. Crittenden, G., 1997. Toxic Niagara, Hazardous Materials Management, December/January 8–20. Environment Canada, 1995. The Research Data Management and Quality Control System (RDMQ) User Manual.
Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Canada. Henry, R.C., Lewis, C.W., Hopke, P.K., Williamson, H.J., 1984. Review of receptor model fundamentals. Atmospheric Environment 18, 1507–1515. Hung, H., Halsall, C.J., Blanchard, P., Li, H.H., Fellin, P., Stern, G., Rosenberg, B., 2001. Are PCBs in the Canadian Arctic Atmosphere Declining? Evidence from 5 years of monitoring. Environmental Science and Technology 35, 1303–1311. IADN Technical Summary of Progress 1990–1996. US/Canada, IADN Scientific Steering Committee, January 1998. Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Canada. Kellerhals, M., Beauchamp, S., Belzer, W., Blanchard, P., Froude, F., Harvey, B., McDonald, K., Pilote, M., Poissant, L., Puckett, K., Schroeder, W.H., Steffen, A., Tordon, R., 2000. Temporal and spatial variability of total gaseous mercury in Canada: preliminary results from the Canadian atmospheric mercury measurement network (CAMNet). In: 11th Annual International Conference on Heavy Metals in the Environment (J. Nriagu, Editor), #1001. University of Michigan, Ann Harbour, MI. Lee, D.S., Dollard, G.J., Pepler, S., 1998. Gas-phase mercury in the atmosphere of the United Kingdom. Atmospheric Environment 32 (5), 855–864. Nakazawa, T., Ishizawa, M., Higuchi, K., Trivett, N.B.A., 1997. Two curve fitting methods applied to CO2 flask data. Environmetrics 8, 197–218. Olson, M.P., Oikawa, K.K., Macafee, A.W., AES Report LRTAP 78-4, 1978, Downsview, Ontario, Canada. Pilgrim, W., Poissant, L., Trip, L., 2000. The Northeast States and Eastern Canadian Provinces mercury study: a framework for action: summary of the Canadian chapter. The Science of the Total Environment 261, 177–184. Poissant, L., 1997. Field observations of total gaseous mercury behaviour: interactions with ozone concentration and water vapour mixing ratio in air at a rural site. Water, Air and Soil Pollution 97, 341–353. Poissant, L., Amyot, M., Pilote, M., Lean, D., 2000. Mercury water–air exchange over the upper St. Lawrence river and Lake Ontario. Environmental Science and Technology 34, 3069–3078. Rotty, R.M., 1987. Estimates of seasonal variation in fossil fuel CO2 emission. Tellus 39B, 184–202. Scheider, W.A., Cox, C., Hayton, A., Hitchin, G., Vaillancourt, A., 1998. Current and temporal trends in concentrations of persistent toxic substances in sport fish and juvenile forage fish in the Canadian waters of the Great Lakes. Environmental Monitoring and Assessment 53, 57–76. Schroeder, W.H, Munthe, J., 1998. Atmospheric mercury-an overview. Atmospheric Environment 32 (5), 809–822. Schroeder, W.H., Keeler, G., Kock, H., Roussel, P., Schneeberger, D., Schaedlich, F., 1995. International field intercomparison of atmospheric mercury measurement methods. Water, Air and Soil Pollution 80, 611–620. Slemr, F., Scheel, H.E., 1998. Trends in atmospheric mercury concentrations at the summit of the Wank mountain, Southern Germany. Atmospheric Environment 32 (5), 845–853.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743 Steffen, A., Schroeder, W.H., 1999. Standard Operating Procedures Manual Procedure for Total Gaseous Mercury Measurements—Canadian Atmospheric Mercury Measurement Network (CAMNet), Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Canada.
3743
Worthy, D.E.J., Levin, I., Trivett, N.B.A., Kuhlmann, A.J., Hopper, J.F., Ernst, M.K., 1998. Seven years of continuous methane observations at a remote boreal site in Ontario, Canada. Journal of Geophysical Research 103, 15995– 16007.
Four years of continuous total gaseous mercury (TGM) measurements at sites in Ontario, Canada P. Blancharda,*, F.A. Froudeb, J.B. Martina, H. Dryfhout-Clarkb, J.T. Woodsc b
a Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Ont., Canada M3H 5T4 Center for Atmospheric Research and Experiment (CARE), Meteorological Service of Canada, R.R. #1, Egbert, Ont., Canada L0L 1N0 c Envirotech Services, 196 Mason St., Bradford, Ont., Canada L3Z 1B2
Received 23 August 2001; accepted 13 May 2002
Abstract Total gaseous mercury (TGM) measurements have been conducted around the Great Lakes since 1997. At two of these sites (Egbert and Burnt Island), TGM concentrations presented significant seasonal variations. Possible explanations for the larger winter–spring/smaller summer–fall concentrations included seasonal meteorological differences, a northern hemispherical increase in coal combustion for wintertime heating, seasonal cycles of atmospheric oxidants and overall hemispherical source–sink relationship. The impact of populated/industrialized areas on the TGM concentrations at rural sites was demonstrated using pollution roses. Trend analyses for Egbert and Point Petre indicated relatively stable TGM concentrations between 1997 and 2000. Principal component analysis of TGM and trace metals confirmed the influence of the industrialized/populated area of Southwestern Ontario on both Egbert and Point Petre as well as metal recovery activities from Northern Ontario. For one episode of significant TGM concentrations, only a general geographical area could be implied as the source of atmospheric mercury. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: TGM; Atmospheric mercury; Trend; PCA; Great lakes
1. Introduction Mercury exists in the atmosphere predominantly in its elemental form (Hg0). Due to its stability, mercury has a residence time of the order of about 1 year and therefore can undergo long range transport (Schroeder and Munthe, 1998). Following deposition, mercury can be converted to methyl mercury, a highly toxic mercury species, which has been found to bio-accumulate in aquatic biota, especially fish. This has resulted in regulatory fish consumption advisories in various parts *Corresponding author. Tel.: +1-416-739-5701; fax: +1416-739-5708. E-mail address: [email protected] (P. Blanchard).
of North America and Europe, including the Great Lakes. Along with PCBs, mercury remains a contaminant of concern in fish in the Canadian waters of the Great Lakes (Scheider et al., 1998). To obtain spatial and temporal distribution of total gaseous mercury (TGM) levels on the Canadian side of the Great Lakes, a measurement campaign was started in 1997 at sites of the Integrated Atmospheric Deposition Network (IADN). The mercury measurements are part of the Canadian Atmospheric Mercury Network (CAMNet). The scope of the network and preliminary results from all sites have been described before (Kellerhals et al., 2000). In this paper we present the results of 4 years of continuous ambient TGM monitoring at two sites and 18 months at another site in Ontario, Canada.
1352-2310/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 2 - 2 3 1 0 ( 0 2 ) 0 0 3 4 4 - 8
3736
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
2. Experimental Total gaseous mercury was measured at Egbert (EGB) (44.22N, 79.78W), a rural site located about 65 km northwest of Toronto, since January 1997; Point Petre (PPT) (43.83N, 77.15W), IADN master station on Lake Ontario, since January 1997 and Burnt Island (BNT) (45.8N, 82.95W), IADN master station on Lake Huron since April 1998 (Fig. 1). At all stations, the measurements are made using Tekrant 2537A mercury vapour analyzers. This instrument has been described elsewhere (Schroeder et al., 1995; Poissant, 1997; Lee et al., 1998). In this paper the operational definition of total gaseous mercury (TGM) for Tekran-based measurements will be used (Schroeder et al., 1995). Briefly, after TGM was collected on a gold cartridge, it was thermally desorbed and detected by cold vapour atomic fluorescence spectroscopy. Two cartridges provided continuous measurements. The integration time was 30 min for 1997, decreased to 15 min in 1998, 1999 and 2000 at all stations. The instrument was calibrated daily using an internal mercury source verified quarterly by manual injections (Steffen and Schroeder, 1999). The data were quality controlled using RDMQt (Environment Canada, 1995) and were hourly averaged. Trace metals data have been collected at the sites as part of IADN since the early 1990s. Every 12 days, composite samples were collected for 24 h using a high volume sampler (PM 10) and subsequently analyzed
using inductively coupled plasma-mass spectrometry (IADN Technical summary, 1998). The available trace metals data for Egbert and Point Petre (1997, 1998) were used in conjunction with the TGM data for principal component analysis (Henry et al., 1984). This statistical method has been described before (Blanchard et al., 1997). Briefly, principal components were extracted from a correlation matrix and rotated by Varimax to facilitate interpretation. Components with eigenvalues greater than one were retained. This agreed with other criteria such as scree plot and analysis of variance. All computational calculations were conducted using SAS statistical software (version 6.12). For this analysis, TGM data were averaged to correspond to the daily sampling period of the trace metals measurements.
3. Results and discussion 3.1. Annual variability and co-pollutants Monthly TGM and temperature means are presented for the three sites in Figs. 2a–c for Egbert, Burnt Island and Point Petre, respectively. Similar seasonal variations are evident for the sites of Egbert and Burnt Island with a maximum in February–March and a minimum in August–September. (For both sites, air temperatures were significantly anti-correlated with TGM (R ¼ 0:21 for EGB and 0.26 for BNT).) The pattern of higher
N Burnt Island Ontario
Point Petre Lake H ron
Egbert Lake Ontario
Buoy Buffalo
Lake Erie Detroit
Fig. 1. Locations of the sampling sites.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743 4.0
15
2.5
10
2.0 5
1.5 1.0
0
0.5
-5
0.0 12/96 4.0
12/98
06/99
12/99
06/00
12/00 25 20 15
2.5
10
2.0 5
1.5 1.0
0
0.5
-5
0.0 12/96
-10 06/97
12/97
06/98
12/98
06/99
12/99
06/00
12/00 25
Point Petre
3.5
TGM, ng/m3
06/98
3.0
4.0
(c)
12/97
20
3.0
15
2.5
10
2.0 5
1.5 1.0
0
0.5
-5
0.0 12/96
Temperature, °C
TGM, ng/m3
(b)
-10 06/97
Burnt Island
3.5
Temperature, °C
20
3.0
Temperature °C
TGM, ng/m3
(a)
25
Egbert
3.5
3737
-10 06/97
12/97
06/98
12/98
06/99
12/99
06/00
12/00
Fig. 2. Monthly TGM and temperature means for (a) Egbert, (b) Burnt Island and (c) Point Petre. The boxes represent the 25th and 75th percentiles. The thick line is the mean and the fine line is the median. The error bars are the 10th and 90th percentiles.
Ozone, ppb 0
20
40
60
80
100
2.2
TGM, ng/m
3
2.0
1.8
1.6
1.4
PM2.5 Ozone SO2
1.2
1.0 0
10
20
30
40
PM2.5, SO2, µg/m
3
Fig. 3. TGM relationships with SO2, PM2.5 and Ozone for Egbert, 1997–2000.
3738
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
mercury concentrations in winter and decreasing in summer–fall has been observed before (Slemr and Scheel, 1998; Ames et al., 1998). Several factors might contribute to this behaviour, they include differences in meteorological conditions between summer and winter such as reduced mixing heights and higher wind speeds in winter, larger removal from the atmosphere by wet and dry deposition during warmer months. One other contributing factor might be the northern hemispherical wintertime increase in coal combustion for domestic heating purposes (Rotty, 1987). This is supported by the relationship seen at Egbert between TGM and daily sulfur dioxide (SO2) as shown in Fig. 3. Each point represents the average of 100 daily measurements. The agreement between TGM and SO2 (R2 ¼ 0:85 for a logarithmic regression) at Egbert likely results from similar sources, e.g. coal combustion for these two species. Lee et al. (1998) used continuous SO2 measurements as a tracer of local combustion sources in their examination of TGM concentrations in the UK. The seasonal cycle of atmospheric oxidant (e.g. ozone) might be also partly responsible for the seasonality of TGM at Egbert and Burnt Island although for ozone concentration X12 ppb (Fig. 3) there is no relationship between ozone and TGM. Segregation using water vapour mixing ratio (Poissant, 1997) was attempted but no significant correlation was obtained. However below 12 ppb, there is a negative relationship between TGM and ozone. This is likely due to pollution episodes originating from the industrialized/populated area of Toronto to the south–southeast of Egbert. Indeed nitric oxide data correlate well with TGM data (not shown), in effect titrating the ozone present in the anthropogenic plume reaching Egbert. The influence of the anthropogenic activities of Southern Ontario as sources of pollutants reaching Egbert is further established from the relationship between TGM and PM2.5 (Fig. 3) where, for this comparison, each point represents 1000 hourly measurements. In ‘‘clean air’’ situations (low levels of PM2.5) there is a lack of correlation due to background TGM concentrations. On the other hand, for PM2.5 concentrations larger than 15 mm/m3, other sources of PM2.5 unrelated to TGM gain importance, e.g. secondary formation of PM2.5. The result of this comparison is interesting since it implies that the correlation between TGM and PM2.5 could be used to indicate secondary production of aerosols. Finally the seasonal cycle at Egbert and Burnt Island might also be influenced by the behaviour of TGM on a hemispherical scale. Recent data of TGM vertical profiles (Banic et al., 2002) indicate that the Arctic springtime TGM depletion episodes could result in a significant hemispherical sink of TGM annually. The observations for BNT and EGB reflect this northern hemispheric behaviour of ‘‘background’’ TGM concentrations with a fairly strong seasonal cycle, albeit
influenced by local anthropogenic sources as discussed above. This is displayed in Figs. 4a–c where pollution roses are plotted for EGB, BNT and PPT respectively. Each bin represents 0.5 ng/m3 TGM, 301 wind direction with colours indicating frequency (%). Each bin has been normalized for overall wind direction frequency.
Fig. 4. TGM pollution roses for (a) Egbert, (b) Burnt Island and (c) Point Petre. Numbers in white are TGM concentrations in ng/m3.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
3739
At Egbert (Fig. 4a), episodes of larger TGM concentrations come from the industrialized/populated areas to the south–southeast of the site, as discussed above from the analysis of co-pollutants. The rose also shows that the lowest TGM concentrations are predominantly from the NNW which is consistent with clean air masses from Northern Canada. For Burnt Island (Fig. 4b), larger TGM concentrations are associated with air masses from the east to southeast again pointing to populated areas of Southern Ontario. TGM observations at Point Petre present a similar seasonal cycle (Fig. 2c) to Egbert and Burnt Island except that larger concentrations are superimposed to the annual cycles, particularly in summer 1998 and 1999. The pollution rose (Fig. 4c) points to the Southerly sectors as sources of TGM episodes for that site. Anthropogenic activities in New York state and Southwestern Ontario are likely the largest contributor to the TGM episodes observed at Point Petre but volatilization from Lake Ontario could also be of some importance (Poissant et al., 2000).
conducted for these data using a digital filter technique described by Nakazawa et al. (1997). Briefly, a long-term trend and seasonal cycle were determined by fitting a smoothing cubic spline and Fourier components to the data. The digital filter method has been used successfully to assess trends of greenhouse gases CO2 and CH4 (Worthy et al., 1998) as well as persistent organic pollutants such as PCB (Hung et al., 2001). Long and short term variations of the trend and the seasonal cycle were extracted using two cutoff periods: a short term cutoff period set to 4 months and a long term cutoff period set to 24 months. The results (Figs. 5a and b) indicate that for both sites, there is no clear trend for TGM between 1997 and 2000. In contrast, Slemr and Scheel (1998) found a decreasing atmospheric mercury trend between 1990 and 1996 for a site in Germany associated with large changes in regional emissions over that period. Although Canadian emissions decreased significantly between 1990 and 1995, current emissions have remained somewhat steady around 10 t/yr (Pilgrim et al., 2000).
3.2. Trend analyses
3.3. Source–receptor relationship
Four complete years of data were available for the sites of Egbert and Point Petre. Trend analyses were
Throughout the period, measurements of several trace metals have been conducted at the three sites as part of
4.0
Egbert
3.5
TGM, ng/m
3
3.0 2.5 2.0 1.5 1.0 0.5 0.0 1997
1998
(a)
1999
2000
2001
Year 4.0
Point Petre
TGM, ng/m
3
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 1997
(b)
1998
1999
2000
2001
Year
Fig. 5. Trends and seasonal cycles of TGM for (a) Egbert and (b) Point Petre: — trend, – seasonal cycle, + TGM.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
3740
trajectories for general anthropogenic/coal factor were similar. In the case of Egbert (Table 2), TGM is associated with two components (PC2 and PC4). Component 2 represent general anthropogenic activities and associated back-trajectories point again to the industrialized, populated areas of Southern Ontario. The other component containing a small loading of mercury (PC4) is associated with Co, Cr, Cu and Ni. Two events are responsible for this component and back-trajectories indicate air flowing from the northwest of the site passing over the Sudbury area where production of nickel and copper from sulphides ores mined in the area constitutes a significant source of atmospheric emission.
IADN. Results were available for 1997–1998 for Point Petre and Egbert. Principal component analysis was used to try and identify potential sources of metals to the two sites. This was done in conjunction with air parcel back-trajectories (Olson et al., 1978) incorporating geographical information to the statistical analysis. The results of the PCA analysis are presented in Tables 1 and 2 where loadings o0.3 have been omitted. The results of the analysis for Point Petre are similar to what has been obtained previously (Blanchard et al., 1997). The results include a crustal component (PC1), general anthropogenic/oil combustion (PC2), smelter (PC4) and another general anthropogenic/coal combustion component (PC3). The latter is particularly interesting since it has the higher loading for TGM associated with Ag, Cd, Cr, Mn, Mo, Se, Sn and Zn. For this component, three days Lagrangian back-trajectories associated with the highest 10% of scores were calculated corresponding to the mid-point of the sampling period (Fig. 6). These indicate air parcels passing over the heavily populated and industrialized areas of the lower Great Lakes to the southwest of the site. Factor analysis has been used previously on metals data combined with particulate mercury data (PM 2.5 and 10) from Perch River, NY in 1992–1994 (Ames et al., 1998). These authors found similar results to this analysis with industrial/coal combustion and smelter sources. Further, their back
Although the TGM concentrations varied only slightly over the years, some short term regional episodes of larger concentration deserve scrutiny. To discuss one particular episode, data from another site are needed. TGM levels from a buoy (43.24N, 79.26W) outfitted with a Tekran reached 70–80 ng/m3 on October 1–2, 2000 (Fig. 7). At Egbert (97 km north of the buoy), TGM concentrations reached 26 ng/m3 and were seen on the night of 2 October Point Petre (186 km northwest of
Table 1 Varimax rotated principal components analysis of 1997–1998 Point Petre trace elements and TGM data
Table 2 Varimax rotated principal components analysis of 1997–1998 Egbert trace elements and TGM data
PC1 Ag Al As Ba Cd Co Cr Cu Fe Mn Mo Ni Pb Sb Se Sr Sn Ti V Zn Hg Eigenvalues %Variance explained
PC2
PC3
PC4
0.48 0.92 0.65
0.36
0.31
0.41
0.65 0.87
0.88 0.78 0.41
0.32 0.39 0.85
0.41 0.69 0.92 0.82
0.33 0.83
0.79 0.50
0.76 0.35 0.34 8.5 40.6
0.81 0.38 (0.14) 3.2 15.4
0.35
0.67 0.73 1.7 8.0
0.54
1.5 7.0
PC1
PC5
0.90
0.66 0.84
3.4. Episode of significant TGM concentration
1.1 5.3
Ag Al As Ba Cd Co Cr Cu Fe Mn Mo Ni Pb Sb Se Sr Sn Ti V Zn Hg Eigenvalues % Variance explained
PC2
PC3
PC4
0.72 0.97 0.73 0.72
0.56 0.33
0.64
0.67 0.30 0.45
0.50 0.71 0.69
0.96 0.94 0.85 0.58 0.78 0.67
0.34 0.68 0.32
0.86
0.95 0.76 0.97 0.46 9.3 44.2
0.60 0.74 0.66 3.3 15.5
0.33 2.1 10.1
(0.25) 1.3 6.0
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
3741
Ontario
Point Petre
Fig. 6. Back-trajectories (950 mb) at Point Petre for the 10% of samples with the highest scores for PC3. Circles: 21 September 1998; squares: 17 June 1998; triangles: 28 July 1997 and crosses: 16 August 1998.
80 Egbert Point Petre Buoy
70
TGM, ng/m
3
60 50 40 30 20 10 0 0:00:00 Sept 30
0:00:00
0:00:00
0:00:00
Oct. 1
Oct. 2
Oct. 3
0:00:00
Fig. 7. TGM concentrations at the Lake Ontario buoy, Egbert and Point Petre during the October 2000 episode.
the buoy) experienced an increase in concentration simultaneously to the buoy, although the concentrations observed were much lower (peaking around 6 ng/m3). Wind direction was from the south with averaged speed 4–5 m/s. Concentrations for other gases such as CO, NO2, O3 and even PM2.5 remain relatively constant
throughout the period (J. Brook, personal communication). Back trajectories indicate that the air mass traveled over the Niagara area, eastern tip of Lake Erie, Buffalo area and Western Pennsylvania. The discrete nature of the TGM peak points to localized sources. Considering the industrialized nature of the Niagara
3742
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743
region (Crittenden, 1997), it is foreseeable that the high levels of TGM observed at the buoy originated from that area.
4. Conclusion For two sites in Ontario, the seasonal pattern of total gaseous mercury was indicative of background northern hemispherical conditions superimposed with regionally influenced episodes. The examination of co-pollutants such as SO2, O3 and PM2.5 helped to assess whether seasonal patterns were due to factors such as winter coal combustion, seasonality of atmospheric oxidant and northern hemispheric source–sink relationship. No clear trend was apparent for the years of measurements in this study (1997–2000). Principal component analysis of TGM and trace metals data pointed to populated/ industrialized areas of Southern Ontario and metals recovery operations of Northern Ontario as sources of TGM episodes. The advantage of the short time resolution afforded by the continuous TGM analyzer used in this study was displayed during a significant episode of TGM concentration.
Acknowledgements The authors would like to thank the site operators D. Smith, F. Orford and particularly C. Green. C. Audette for the IADN trace metals data and D. Worthy for the use of the digital filter model. Dr. J.R. Brook for copollutants data from Toronto. Drs. K Puckett and C. Banic for helpful discussions.
References Ames, M., Gullu, G., Olmez, I., 1998. Atmospheric mercury in the vapor phase and in fine and coarse particulate matter at Perch River, New York. Atmospheric Environment 32 (5), 865–872. Banic, C.M., Beauchamp, S.T., Tordon, R.J., Schroeder, W.H., Steffen, A., Anlauf, K.A., Wong, H.K.T., 2002. Vertical distribution of gaseous elemental mercury in Canada. Journal of Geophysical Research, submitted for publication. Blanchard, P., Hopper, J.F., Hoff, R.M., 1997. Principal component analysis of IADN data from a site in Ontario. 97-A805, Presented at the Air & Waste Management Association’s 90th Annual Meeting & Exhibition, June 8– 13, 1997, Toronto, Ont., Canada. Crittenden, G., 1997. Toxic Niagara, Hazardous Materials Management, December/January 8–20. Environment Canada, 1995. The Research Data Management and Quality Control System (RDMQ) User Manual.
Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Canada. Henry, R.C., Lewis, C.W., Hopke, P.K., Williamson, H.J., 1984. Review of receptor model fundamentals. Atmospheric Environment 18, 1507–1515. Hung, H., Halsall, C.J., Blanchard, P., Li, H.H., Fellin, P., Stern, G., Rosenberg, B., 2001. Are PCBs in the Canadian Arctic Atmosphere Declining? Evidence from 5 years of monitoring. Environmental Science and Technology 35, 1303–1311. IADN Technical Summary of Progress 1990–1996. US/Canada, IADN Scientific Steering Committee, January 1998. Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Canada. Kellerhals, M., Beauchamp, S., Belzer, W., Blanchard, P., Froude, F., Harvey, B., McDonald, K., Pilote, M., Poissant, L., Puckett, K., Schroeder, W.H., Steffen, A., Tordon, R., 2000. Temporal and spatial variability of total gaseous mercury in Canada: preliminary results from the Canadian atmospheric mercury measurement network (CAMNet). In: 11th Annual International Conference on Heavy Metals in the Environment (J. Nriagu, Editor), #1001. University of Michigan, Ann Harbour, MI. Lee, D.S., Dollard, G.J., Pepler, S., 1998. Gas-phase mercury in the atmosphere of the United Kingdom. Atmospheric Environment 32 (5), 855–864. Nakazawa, T., Ishizawa, M., Higuchi, K., Trivett, N.B.A., 1997. Two curve fitting methods applied to CO2 flask data. Environmetrics 8, 197–218. Olson, M.P., Oikawa, K.K., Macafee, A.W., AES Report LRTAP 78-4, 1978, Downsview, Ontario, Canada. Pilgrim, W., Poissant, L., Trip, L., 2000. The Northeast States and Eastern Canadian Provinces mercury study: a framework for action: summary of the Canadian chapter. The Science of the Total Environment 261, 177–184. Poissant, L., 1997. Field observations of total gaseous mercury behaviour: interactions with ozone concentration and water vapour mixing ratio in air at a rural site. Water, Air and Soil Pollution 97, 341–353. Poissant, L., Amyot, M., Pilote, M., Lean, D., 2000. Mercury water–air exchange over the upper St. Lawrence river and Lake Ontario. Environmental Science and Technology 34, 3069–3078. Rotty, R.M., 1987. Estimates of seasonal variation in fossil fuel CO2 emission. Tellus 39B, 184–202. Scheider, W.A., Cox, C., Hayton, A., Hitchin, G., Vaillancourt, A., 1998. Current and temporal trends in concentrations of persistent toxic substances in sport fish and juvenile forage fish in the Canadian waters of the Great Lakes. Environmental Monitoring and Assessment 53, 57–76. Schroeder, W.H, Munthe, J., 1998. Atmospheric mercury-an overview. Atmospheric Environment 32 (5), 809–822. Schroeder, W.H., Keeler, G., Kock, H., Roussel, P., Schneeberger, D., Schaedlich, F., 1995. International field intercomparison of atmospheric mercury measurement methods. Water, Air and Soil Pollution 80, 611–620. Slemr, F., Scheel, H.E., 1998. Trends in atmospheric mercury concentrations at the summit of the Wank mountain, Southern Germany. Atmospheric Environment 32 (5), 845–853.
P. Blanchard et al. / Atmospheric Environment 36 (2002) 3735–3743 Steffen, A., Schroeder, W.H., 1999. Standard Operating Procedures Manual Procedure for Total Gaseous Mercury Measurements—Canadian Atmospheric Mercury Measurement Network (CAMNet), Meteorological Service of Canada, 4905 Dufferin Street, Toronto, Canada.
3743
Worthy, D.E.J., Levin, I., Trivett, N.B.A., Kuhlmann, A.J., Hopper, J.F., Ernst, M.K., 1998. Seven years of continuous methane observations at a remote boreal site in Ontario, Canada. Journal of Geophysical Research 103, 15995– 16007.
