Relationships between atmospheric transport ... - ACS Publications

Subscriber access provided by Queen Mary, University of London

Article

Relationships between atmospheric transport regimes and PCB concentrations in air at Zeppelin, Spitsbergen Sandy Ubl, Martin Scheringer, and Konrad Hungerbühler Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b02571 • Publication Date (Web): 18 Jul 2017 Downloaded from http://pubs.acs.org on August 1, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26

Environmental Science & Technology

Relationships between atmospheric transport regimes and PCB concentrations in air at Zeppelin, Spitsbergen 1

Sandy Ubl,† Martin Scheringer,∗,†,‡ and Konrad Hungerbühler† †Institute for Chemical and Bioengineering, ETH Zürich, 8092 Zürich, Switzerland ‡RECETOX, Masaryk University, 625 00 Brno, Czech Republic E-mail: [email protected]

2

Abstract

3

Polychlorinated biphenyls (PCBs) are persistent hazardous chemicals that are still

4

detected in the atmosphere and other environmental media although their production

5

has been banned for several decades. At the long-term monitoring site, Zeppelin at

6

Spitsbergen, different PCB congeners have been continuously measured for more than

7

a decade. However, it is not clear what factors determine the seasonal and inter-annual

8

variability of different (lighter vs. heavier) PCB congeners. To investigate the influence

9

of atmospheric transport patterns on PCB-28 and PCB-101 concentrations at Zeppelin,

10

we applied the Lagrangian Particle Dispersion Model FLEXPART and calculated “foot-

11

prints” that indicate the potential source regions of air arriving at Zeppelin. By means

12

of a cluster analysis, we assigned groups of similar footprints to different transport

13

regimes and analyzed the PCB concentrations according to the transport regimes. The

14

concentrations of both PCB congeners are affected by the different transport regimes.

15

For PCB-101, primarily the origin of air masses from the European continent is related

16

to high concentrations; elevated PCB-101 concentrations in winter can be explained by

1 ACS Paragon Plus Environment


17

the high frequency of this transport regime in winter whereas PCB-101 concentrations

18

are low when air is arriving from the oceans. For PCB-28, in contrast, concentrations

19

are high during summer when air is mainly arriving from the oceans, but low when

20

air is arriving from the continents. The most likely explanation of this finding is that

21

local emissions of PCB-28 mask the effect of long-range transport and determine the

22

concentrations measured at Zeppelin.

23

Introduction

24

Polychlorinated biphenyls (PCBs) are a group of environmentally hazardous chemicals be-

25

cause of their high persistence, bioaccumulation potential, and toxicity 1 . They are included

26

in the Stockholm Convention on Persistent Organic Pollutants (POPs), a global treaty with

27

the aim of protecting human health and the environment from hazardous, highly persistent

28

chemicals 2 .

29

PCBs have been widely used as dielectric and coolant fluids, as transformer and capacitor

30

insulating materials, in paints, and as sealing materials 3 . The majority of PCBs have been

31

used in the Northern Hemisphere, mostly between 30◦ and 60◦ N. Due to the remoteness of

32

the Arctic (i. e., the region north of 60◦ N), it is generally assumed that this region received no

33

direct input of POPs. However, many POPs were detected in various environmental media

34

in the Arctic, such as sediments, soils, and the atmosphere 2 . The atmospheric concentrations

35

of these chemicals have been measured routinely within monitoring programs such as the

36

European Monitoring and Evaluation Programme (EMEP) 2 for more than a decade.

37

The main source of PCBs detected in the Arctic results from long-range atmospheric

38

transport 2 in combination with dry/wet deposition to the surface, and transport via ocean

39

currents 4 . Exceptionally high PCB concentrations at Zeppelin have been linked to boreal

40

forest fires 5 and it has been shown that the 20% highest PCB-28 and PCB-101 concentrations

41

are associated with air masses arriving from regions over Europe with known PCB emission

42

sources 6 . 2 ACS Paragon Plus Environment

Page 2 of 26

Page 3 of 26


43

However, temporal trends in PCB concentrations measured in the atmosphere at Zeppelin

44

are complex and difficult to explain. A downward trend until 2004 followed by increasing

45

concentrations has been reported for many PCB congeners 7,8 , but different PCB congeners

46

exhibit different seasonal variability. Lower-chlorinated PCBs show the highest concentra-

47

tions in summer, whereas elevated concentrations of medium-chlorinated PCBs are observed

48

in winter 6 . To identify potential sources and source regions of PCBs and to link PCB con-

49

centrations to long-range transport phenomena, it is essential to understand these complex

50

seasonal trends for different PCBs observed at Zeppelin. Therefore, the aims of this study

51

were i) to identify relevant atmospheric transport regimes for air masses arriving at Zep-

52

pelin, ii) to link seasonal concentration patterns of different PCBs to the identified transport

53

regimes, and iii) to evaluate the importance of local PCB sources contributing to the mea-

54

surements at Zeppelin. For this purpose we used the Lagrangian Particle Dispersion Model

55

(LPDM) FLEXPART to simulate the atmospheric transport of PCB-28 and PCB-101 for

56

the years 2000–2012. Based on a cluster analysis, we merged similar footprints into distinct

57

transport regimes and related measured PCB concentrations to different transport regimes.

58

We compared the seasonal trends of the PCB concentrations measured at Zeppelin to the

59

trends at Andøya, another Arctic measurement site, to assess the influence of local PCB

60

sources and to explain the seasonal PCB trends observed at Zeppelin.

61

Material and Methods

62

Measurement site

63

The Zeppelin observatory is located in the Arctic on Zeppelin Mountain, Spitsbergen, at

64

79◦ N and 12◦ E at 474 meters above sea level. It is generally assumed that the station is

65

above the local inversion layer most of the time and thus is not influenced by local pollution

66

from the nearby community, e.g. Ny-Ålesund. For PCB measurements, 1000 m3 of air are

67

sampled over 48 h and measured on a weekly basis. A detailed description of the sampling 3 ACS Paragon Plus Environment


68

procedure and the analytically methods is provided by Hung et al. 9 , Bossi et al. 10 , Eckhardt

69

et al. 11 , and available from the UNECE-EMEP website: http://www.emep.int. We selected

70

PCB-28 and PCB-101 for our analysis because both mainly occur in the gas phase and are

71

therefore suitable for modeling with a Lagrangian Particle Model. We considered data from

72

June 2000 to 2012. Before June 2000 the PCB measurements were affected by contamination

73

or samples were taken in Ny-Ålesund and not at Zeppelin mountain 7 and are therefore not

74

representative for a background measurement site.

75

Model description

76

To focus on the role of atmospheric transport regarding PCB concentrations in the Arctic we

77

need to employ a model that describes Arctic and mid-latitude transport accurately. LPDMs

78

have a detailed treatment of atmospheric transport (including turbulence and convection),

79

wet and dry deposition, and include simple chemical transformations such as the removal

80

of POPs by hydroxyl (OH) radicals. Additionally, LPDMs can be run backward in time

81

to identify the source regions of chemicals transported in the atmosphere. The LPDM

82

FLEXPART 12–14 has been used successfully to study atmospheric transport patterns in the

83

Arctic 14 as well as to simulate PCB-28 concentrations at Birkenes 11 . Here, FLEXPART was

84

driven by the European Center for Medium Range Weather Forecast (ECMWF) operational

85

analysis data (input) with 91 vertical levels and a horizontal grid spacing of 1◦ × 1◦ . The

86

model was used in backward mode to calculate so-called footprints (output) for Zeppelin.

87

A footprint indicates the potential source region of a pollutant arriving at Zeppelin and

88

consists of residence times expressed in units of seconds per grid cell. The grid cells have

89

a lateral dimension of 1◦ × 1◦ and a height of 100 m. For the model calculation 48 · 104

90

so-called tracer particles (representing infinitesimally small air parcels 13 ) were released from

91

the Zeppelin grid cell during 24 hours 6,11,15 . The tracer particles were traced backward in

92

time for 22 days. FLEXPART considers horizontal and vertical displacements due to the

93

mean atmospheric flow, turbulent mixing, especially in the atmospheric boundary layer, and 4 ACS Paragon Plus Environment

Page 4 of 26

Page 5 of 26


94

vertical transport due to deep convection. The removal of PCB due to the reaction with OH

95

radicals 16 and due to dry and wet deposition are implemented in the FLEXPART model. A

96

detailed description of these processes and their implementation in the FLEXPART model is

97

given by Seibert and Frank 17 and by Eckhardt et al. 11 . As the partitioning of PCBs between

98

gaseous and particulate phase is not considered in FLEXPART, the model is less suitable

99

for calculating the transport of PCBs that are sorbed strongly to atmospheric particles (log

100

KOA ≈ 10 or higher 18 ). To calculate the losses by dry and wet deposition, Henry’s law

101

constants of 33.1 Pa m3 mol−1 (PCB-28) and 31.4 Pa m3 mol−1 (PCB-101) were used 19 and

102

for the reaction of the PCBs with OH radicals, rate constants of 1.1−12 cm3 s−1 for PCB-28 16

103

and 3.4−13 cm3 s−1 for PCB-101 were used 20 .

104

Cluster analysis

105

For each weekly 48-hour sampling interval we calculated a footprint, resulting in 658 indi-

106

vidual footprints for PCB-28 and PCB-101. To identify the main geographical areas over

107

which air masses arriving at Spitsbergen have been transported and thus may have picked

108

up PCBs, we conducted a cluster analysis to group footprints that represent similar trans-

109

port regimes. For the cluster analysis we used the model domain ranging from 35◦ North to

110

85◦ North and treated the residence times in every grid cell as a variable. To account for

111

the non-normally distributed residence time values, we followed the approach described by

112

Sturm et al. 21 and calculated the absolute distance between the ranks of the residence times

113

according to

di,j =

N X

|rank(τk )i − rank(τk )j |, i, j = 1, ..., 658

(1)

k=1 114

where the residence times are indicated as τ , the index k runs over all grid cells from 1

115

to N ), and i and j refer to the sampling interval. The clustering based on the dissimilarity

116

matrix, di,j was then calculated with the k-medoid method 22 . We varied the number of



Page 6 of 26

117

clusters between 3 and 8 and applied the silhouette technique 22 and visual analysis to select

118

the optimum number of clusters for further analysis. The silhouette value for each point is a

119

measure of how similar that point is to points in its own cluster in comparison to points in

120

other clusters. Thus the average of the silhouette values of the entire data set is a measure of

121

how well the data have been clustered. The average silhouette value showed a local maximum

122

of 0.20 for 5 clusters for PCB-28 and 0.19 for 6 clusters for PCB-101. The number of clusters

123

is a compromise between resolution of different potential source regions (more clusters) and

124

a sufficiently high number of footprints in the individual clusters (fewer clusters).

125

We normalized the cluster footprints according to equation 2, where anorm is the nor-

126

malized footprint of a given cluster, a is the average footprint of this cluster and b is the

127

average of all 658 footprints. This procedure resulted in normalized, average footprints per

128

cluster with values ranging from −2 to +2. Normalized cluster footprints with values above

129

0 indicate areas where the air masses resided longer than on average (average footprint) and

130

clusters with values below 0 represent areas where the air masses resided shorter than on

131

average.

anorm =

a−b (a + b)/2

(2)

132

Results

133

Changes of the concentrations of PCB-28 and PCB-101 from 2000–

134

2012

135

The trends of PCB concentrations measured at Zeppelin are not consistent (Fig.1). For PCB-

136

28, concentrations decrease from the middle of 2000 until 2002. In 2002, the concentrations

137

increase again, especially in summer. In 2003 and 2004 they are less variable, and in 2005

138

they increase again and remain on a higher level until the middle of 2009. This structural

139

change of the PCB-28 concentration trend in 2005 was also identified by Wöhrnschimmel 6 ACS Paragon Plus Environment

Page 7 of 26


PCB-28

pg/m 3

15

10

5

0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

PCB-101

pg/m 3

1.5 1 0.5 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Figure 1: Time series of PCB-28 and PCB-101 concentrations measured at Zeppelin in 2000– 2012 (source: http://www.emep.int). Black line: raw data; red line: seasonal trend of the PCB concentrations, calculated by the robust locally weighted scatterplot smoothing (rloess) in Matlab. The ticks are set at the beginning of each year. 140

et al. 8 . In 2010, another change with decreasing PCB-28 concentrations is visible in the data

141

(Fig. 1 top); this change has been identified as a step change by Zhao et al. 23 .

142

For PCB-101, the concentrations show a clear seasonal variability with a winter maxi-

143

mum 6 . Generally, PCB-101 concentrations slightly decrease from the middle of 2000 until

144

2005. In 2006 and 2007, the concentrations slightly increase before they start to decrease

145

again in 2008. This change in PCB-101 concentrations has also been identified by Zhao et

146

al. 23 .



147

Transport regimes

148

The silhouette technique yielded 5 clusters for PCB-28 (Fig. 2) and 6 clusters for PCB-101

149

(Fig. 3), reflecting transport regimes with different origins of air masses arriving at Zeppelin.

150

The reason for the different numbers of clusters is the differences between the footprints for

151

PCB-28 and PCB-101, reflecting the chemicals’ different physicochemical properties. For

152

PCB-28 the footprints of cluster 1 represent air masses that arrive from all directions, with

153

residence times slightly enhanced over Russia. Footprints associated with clusters 2 and 5

154

represent air originating from the Arctic ocean (cluster 2) and Atlantic ocean (cluster 5). The

155

footprints of cluster 3 represent air masses coming from Europe. The footprints of cluster 4

156

represent air masses with longer-than-average residence times predominantly over Canada,

157

the US and the Pacific ocean. Like PCB-28, the footprints for PCB-101 forming cluster 1

158

have average residence times over Europe, the US and the oceans. Air masses dominantly

159

originating from the oceans are reflected in the footprints of cluster 2 (Arctic ocean) and

160

cluster 3 (Atlantic ocean). Air masses from the continents are mainly represented by cluster

161

4 (Europe, Russia, eastern Asia) and cluster 5 (Europe, Russia and US). The footprints of

162

cluster 6 represent air masses originating from the Pacific ocean but also from Canada and

163

eastern Asia.

164

165

Similar transport regimes were also described by Hirdman et al. 15 and are consistent with general transport pathways to the Arctic 14,24 .

166

To evaluate the seasonal dependence of the transport regimes, the total and seasonal foot-

167

print frequencies per cluster were calculated (Table 1). Ocean transport regimes represented

168

by footprints in clusters 2 and 5 for PCB-28 and clusters 2 and 3 for PCB-101 dominantly oc-

169

cur during the summer months, whereas the footprints representing transport regimes from

170

the continents (cluster 3 for PCB-28 and clusters 4 and 5 for PCB-101) dominate during the

171

winter months.

172

Transport of air masses from the Pacific ocean (cluster 4 for PCB-28, cluster 6 for PCB-

173

101) predominantly occurs in spring and autumn for both PCBs. Thus, the occurrence 8 ACS Paragon Plus Environment

Page 8 of 26

Page 9 of 26


174

of different transport regimes represented by the footprints of individual clusters strongly

175

depend on the season, which is again in line with the known Arctic transport pattern 14 . Table 1: Frequencies of PCB-28 and PCB-101 footprints per cluster and season in spring (March, April, May), summer (June, July, August), autumn (September, October, November), and winter (December, January, February). cluster PCB-28 1 2 3 4 frequency % 23 24 26 15 all 154 157 174 97 55 16 33 47 spring summer 2 110 0 7 67 31 25 33 autumn winter 30 0 116 10

5 12 76 7 54 14 1

cluster PCB-101 frequency 1 2 3 4 % 18 17 22 13 all 118 113 144 84 spring 32 24 19 29 summer 15 63 80 1 51 20 34 19 autumn winter 20 6 11 35

5 16 109 15 1 22 71

6 14 90 39 1 24 14

176

Measured PCB concentrations in relation to transport regimes

177

For each of the 658 time points there is a modeled footprint and a measured PCB concentra-

178

tion. The transport regimes were derived from the clustering of the footprints, see preceding

179

section, and every individual PCB measurement was assigned to its respective transport

180

regime. We then calculated median PCB concentrations for each transport regime to assess

181

whether the different transport regimes are reflected by distinct PCB concentrations.

182

For PCB-28, highest median concentrations were observed when air masses arrived from

183

the Arctic and Atlantic oceans (corresponding to clusters 2 (2.41 pg/m3 ) and 5 (3.20 pg/m3 )

184

(Fig. 4 left)). Transport of air masses from all directions (cluster 1; 1.58 pg/m3 ), Europe

185

and the US (cluster 3; 1.87 pg/m3 ) and Canada and US (cluster 4; 1.66 pg/m3 ) resulted in 9 ACS Paragon Plus Environment


Cluster: 1

Page 10 of 26

Cluster: 2 1.5

o

1

Cluster: 3

Cluster: 4

0.5

0

-0.5

Cluster: 5

-1

-1.5

Figure 2: Normalized footprints (PCB-28) for 5 different transport regimes for Zeppelin from 2000–2012 as obtained from FLEXPART simulations and footprint clustering. Values greater than 0 indicate areas where the air masses reside longer than on average (average footprint) and clusters with values below 0 indicate areas where the air masses reside shorter than on average. Values above 1 or below −1 indicate 3 times longer or shorter residence times, respectively, than on average. The pink circle indicates the station location. The maps were generated with the M_Map package of Matlab.


Page 11 of 26


Cluster: 1

Cluster: 2 1.5

o

1

Cluster: 3

Cluster: 4

0.5

0

-0.5

Cluster: 5

Cluster: 6

-1

o -1.5

Figure 3: Normalized footprints (PCB-101) for 6 different transport clusters for Zeppelin from 2000–2012 as obtained from FLEXPART simulations and footprint clustering. Values greater than 0 indicate areas where the air masses reside longer than on average (average footprint) and clusters with values below 0 indicate areas where the air masses reside shorter than on average. Values above 1 or below −1 indicate 3 times longer or shorter residence times, respectively, than on average. The pink circle indicates Spitsbergen where the measurement station is located. The maps were generated with the M_Map package of Matlab.



186

similar PCB concentrations that are all lower than those represented by clusters 2 and 5.

187

For PCB-101, we observed the highest median concentration (0.51 pg/m3 , 0.43 pg/m3 )

188

when air masses were transported over Europe (cluster 5) and Russia (cluster 3 (Fig. 4

189

right)). For air masses coming from all directions and the Arctic ocean (clusters 1 and 3) the

190

median concentrations were 0.37 pg/m3 , 0.36 pg/m3 . Lowest median values were associated

191

with air masses transported over Arctic and Pacific ocean (clusters 2 (0.30 pg/m3 ) and 6 (0.32

192

pg/m3 )). Thus, PCB-101 concentrations were high when air arrived from the continents and

193

PCB-28 concentrations were high when air arrived from the oceans. The Wilcoxon rank-

194

sum test was applied to evaluate whether the concentration distribution related to a specific

195

cluster is significantly different from the concentration distribution in another cluster. A

196

significant difference of the PCB concentration distribution between clusters suggests that the

197

identified transport regimes are responsible for the different PCB concentrations measured

198

at Zeppelin.

199

For PCB-28, the concentrations related to transport of air masses from the ocean (clus-

200

ters 2 and 5) are significantly higher than PCB-28 concentrations related to the remaining

201

transport regimes (clusters 1, 3, and 4) when the air is mainly transported over land. For

202

PCB-101, concentrations related to the pure ocean transport regime (cluster 2) are signif-

203

icantly lower in comparison to PCB-101 concentrations related to the remaining transport

204

regimes represented by clusters 1, 3, 4, 5 and 6. The PCB-101 concentrations related to

205

air masses arriving from Europe (cluster 5) and Russia (cluster 4) are significantly higher

206

compared to PCB-101 concentrations related to all other transport regimes represented by

207

clusters 1, 2, 3, and 6.

208

To investigate whether the PCB concentrations within individual transport regimes ex-

209

hibit seasonal variations, we calculated the median PCB concentration per transport regime

210

per season (colored dots in Fig. 4). The larger the spread of the colored dots (corresponding

211

to the seasons), the stronger are the seasonal variations of the PCB concentrations, which in

212

turn can be interpreted as changing emissions in the respective source regions. For PCB-28,


Page 12 of 26

Page 13 of 26


23%

24%

26%

15%

18%

12%

17%

22%

13%

16%

14%

8 PCB-101

6

0.8

pg/m 3

pg/m 3

spring

1

PCB-28

4 2

summer 0.6

autumn

0.4 0.2

winter 0

0 1

2

3

4

5

1

cluster number

2

3

4

5

6

cluster number

Figure 4: Boxplot of PCB-28 (left) and PCB-101 (right) concentrations classified by the transport regimes. Boxes show the interquartile range; whiskers extend to the most extreme values. To focus on the difference between the clusters, outliers are not plotted. The colored dots are the median concentrations per season with spring (MAM), summer (JJA), autumn (SON) and winter (DJF). The percent values on top of the panels indicate the frequencies of the transport regimes. Seasons with fewer than 4 footprints in a given cluster are not shown. 213

the general concentration variability as well as the seasonal variation is highest for the ocean

214

transport regimes (clusters 2 and 5). PCB-28 concentrations related to cluster 2 during the

215

summer season were clearly higher than the median and autumn and spring concentrations

216

were clearly lower than the median. PCB-28 concentrations related to the Atlantic trans-

217

port regime (cluster 5) were below the 25th percentile during spring. For the Arctic-Pacific

218

ocean transport regime (cluster 4) the PCB-28 concentrations were below the 25th percentile

219

during the winter season. The transport from Canada, US and Europe (clusters 3) did not

220

reveal any seasonal variations.

221

For PCB-101, a considerable seasonal variability of concentrations was observed for air

222

masses arriving from all directions (cluster 1) and from the Atlantic ocean (cluster 3) with

223

concentrations during summer (cluster 1) and winter (cluster 3) above the 75th percentile.

224

During spring PCB-101 concentrations were above the median for transport regimes repre-

225

sented by clusters 1, 4, 5 and 6.

226

The time series of PCB-28 and PCB-101 (Fig. 1) do not show a clear trend and, therefore, 13 ACS Paragon Plus Environment


cluster

pg/m 3

4

PCB-28

1 2 3 4 5

3 2 1 2000

2002

2004

2006

2008 cluster

pg/m 3

0.6

PCB-101

0.4 0.2 2000

2002

2004

2006

1 2 3 4 5 6

2008

Figure 5: 4-year running median per cluster of measured PCB-28 (top) and PCB-101 (bottom) concentrations. The different clusters are color coded; the numbers in the legend correspond to the cluster numbers in Figures 2 and 3.


Page 14 of 26

Page 15 of 26


227

we did not subtract a background to correct the data. To evaluate whether the median

228

PCB concentration per cluster, considering the years from 2000–2012, is dominated by high

229

PCB concentrations in individual years, we calculated the running median over 4 years for

230

the measured PCB concentration per cluster (Fig. 5). For PCB-28, a decreasing trend of

231

the running medians for all the 5 clusters was observed for the years 2000–2002. From

232

2002 to 2005 the median concentration increased and then decreased again after 2006. The

233

results from the 4-year running medians again indicated that the highest concentrations were

234

associated with air masses arriving from the oceans (clusters 2 and 5), in agreement with the

235

results from the whole time series. In recent years, the difference of the PCB concentrations

236

resulting from the distinct transport regimes became less prominent. For PCB-101, running

237

medians decreased for the ocean transport regimes (cluster 2 and 3). The running medians

238

for the two transport regimes receiving air from Russia and eastern Asia (cluster 4 and 6)

239

are invariant until 2005 and decrease afterwards. For the European transport regime and

240

the all-directions transport regime the running medians decreased from 2000 to 2003 and

241

stayed on a similar level until 2007 when the concentrations start to decrease again.

242

The highest PCB-101 concentrations were associated with air masses arriving from Eu-

243

rope (cluster 5) and Russia (cluster 4) in agreement with the results from the whole time

244

series.

245

PCB concentrations versus residence time over land

246

The different PCB concentrations associated with the distinct transport regimes can be

247

related to (i) varying residence times of the respective air masses in the respective source

248

region and/or (ii) the variable PCB emission rates within the source regions. To disentangle

249

these two effects, we evaluated the median PCB concentrations observed for the different

250

transport regimes as a function of the mean residence time over land and over ocean (Fig.

251

6). The mean residence times over land and ocean were calculated by splitting each footprint

252

into a land and an ocean fraction and then summing and normalizing the respective footprint 15 ACS Paragon Plus Environment


5

median PCB-101 [pg/m3 ]

median PCB-28 [pg/m3 ]

5 3 R 2 = 0.65

R 2 = 0.11

2

2

2

3 4

3 4 1

1

1 0

10

20

30

Page 16 of 26

55

0.5 R 2 = 0.66

4

4

0.4 1

3

31

6 0.3

6

2

2 2

R = 0.71 0.2

40

10

mean surface residence time [s]

20

30

40

mean surface residence time [s]

Figure 6: Median of concentrations of PCB-28 (left) and PCB-101 (right) of the different transport regimes (given by the cluster numbers) as a function of mean residence time over land (black) and over the ocean (blue) per transport regime. 253

fractions per cluster. For PCB-101, the measured concentrations were positively correlated

254

with the mean surface residence times over land (R2 = 0.66) and negatively correlated with

255

mean surface residence times over oceans (R2 = 0.71). Thus, the longer the air masses reside

256

close to the surface over land (high PCB emissions, see Breivik et al. 25 ,26 ), the higher are the

257

PCB-101 concentrations and, vice versa, the longer the air resided over the oceans, where

258

PCB emissions are low, the lower are the PCB-101 concentrations.

259

For PCB-28, in contrast, the measured concentrations were negatively correlated with

260

the mean surface residence time over land (R2 = 0.65) and over the ocean (R2 = 0.11). The

261

negative correlation between residence times over land and the measured PCB-28 concen-

262

trations reveals that the main primary PCB-28 sources do not have a strong influence on

263

PCB-28 concentrations measured at Zeppelin. For the ocean, there is no clear relationship;

264

this finding will be discussed in the next section.


Page 17 of 26


265

Discussion

266

The seasonal trends of PCB-101 can be well explained by dominant transport regimes de-

267

livering air masses from distinct source regions to the Arctic, in combination with primary

268

PCB emissions from the respective source regions 3 . High residence times over land coincide

269

with high PCB concentrations measured at Zeppelin.

270

In contrast, PCB-28 concentrations are high when residence times over the main primary

271

source regions (e.g. Europe) are low and thus, the seasonal trends of the PCB concentrations

272

at Zeppelin can neither be explained by the transport regimes nor by the primary PCB

273

emissions. Thus, alternative emission pathways such as re-emissions of PCB from the ocean

274

water and ice or local PCB sources which both may account for the observed discrepancies

275

are discussed in the following sections.

276

Re-emission of PCB from ocean water and ice

277

Air transport from the Arctic and Atlantic ocean (clusters 2 and 5) mainly occurs during

278

summer and thus additional PCB emissions in summer may be responsible for the elevated

279

PCB concentrations. Re-emissions of PCBs from the ocean water and from sea ice have been

280

considered as potential sources of POPs 7,27 . Net volatilization of PCBs to the atmosphere

281

have been reported for the tropical Atlantic 28,29 . However, the potential source regions or

282

the air masses reaching Zeppelin during summer (clusters 2 and 5) are within the Arctic

283

ocean for which net deposition dominates 30,31 . Friedman and Selin 32 also exclude secondary

284

PCB emissions as explanation for elevated PCB concentrations in summer, because sec-

285

ondary emissions would have to be unrealistically high to explain the seasonal discrepancies

286

between measured and modeled PCB. Temperature-dependent changes in the surface area

287

and structure of snow and ice may affect deposition and re-emission patterns of PCBs 33,34 ,

288

but the extent to which these processes may influence the PCB concentrations in the Arctic

289

remains unclear. However, the sharp decrease in the PCB concentrations observed after 2010



290

is inconsistent with such an emission pathway.

291

Local PCB sources

292

A collaborative project (2007–2010) between the Geological Survey of Norway (NGU), the

293

Governor of Spitsbergen and the Climate and Pollution Agency (Klif) (2007–2010) aimed

294

to identify and eliminate local PCB sources in Spitsbergen. Sources of PCBs were detected

295

on land in Barentsburg and Pyramiden 4,35 . Furthermore, elevated PCB concentrations were

296

detected in soils and PCBs were found in paints, in addition to other PCB sources not further

297

listed 4 . The use of PCB-containing fluids in underground mining (especially coal mining at,

298

e.g., Barentsburg) may additionally contribute to local PCB emissions close to Zeppelin, but

299

no quantitative data are available on the amounts of PCB released into the environment

300

by ventilation systems, mine output and pit water 2 , which makes it very challenging to

301

quantitatively account for potential local PCB sources. However, during summer, elevated

302

temperatures favor the emission of volatile PCBs and the atmospheric boundary layer height

303

is above the elevation of the Zeppelin mountain 36,37 . Both factors support the hypothesis

304

that local PCB emissions are responsible for the high PCB concentrations observed during

305

the summer season when air masses dominantly arrive from the oceans and residence times

306

are high close to the station (clusters 2 and 5). This hypothesis is also supported by Fried-

307

man and Selin 32 , who consider local primary emissions a more probable driver for summer

308

concentration maxima compared to re-emissions from surface media because of the greater

309

effect of temperature changes on primary volatilization emissions compared to secondary

310

emissions. Furthermore, the PCB concentrations measured at Zeppelin are substantially

311

higher than the PCB concentrations reported from six other Arctic monitoring stations 4 .

312

Local PCBs sources were removed from Svalbard after 2009 4 , which may explain the sharp

313

decrease of the PCB concentrations observed between 2009 and 2010, but the presence of

314

additional local emissions, such as heavily contaminated soils, likely still persists.


Page 18 of 26

Page 19 of 26


315

Comparison of PCB concentrations measured at Spitsbergen and

316

Andøya

317

To evaluate whether local PCB emissions or emissions from ice and open water were dominat-

318

ing the PCB measurements during summer in Spitsbergen, we compared the PCB concentra-

319

tions from Spitsbergen and Andøya (data available since 2009), another Arctic measurement

320

site located 10 degrees south of Zeppelin. PCB-28 concentrations at Zeppelin are consider-

321

ably higher, mostly by more than a factor of 3 compared to Andøya. After 2009, however,

322

the PCB-28 concentrations at Zeppelin and Andøya differ only by a factor of 2. Both stations

323

receive air from the ocean, indicating that re-emissions of PCBs from the open ocean and or

324

from sea ice would influence both stations equally. The time series of PCB-28 measured at

325

Andøya (see Fig. 12 38 ) clearly shows a seasonal cycle with higher concentrations in winter

326

than in summer, consistent with the Arctic transport regime. However, PCB-28 concentra-

327

tions at Zeppelin show an opposite seasonal cycle with elevated concentrations in summer,

328

suggesting that PCB concentrations measured at Zeppelin especially before 2009 and their

329

seasonal variations were dominated by local emissions on Spitsbergen.

330

Environmental Significance

331

For PCB-101, we are able to explain 66% of the concentration variability between the trans-

332

port regimes by the residence time over land (Figure 6). However, PCB-28 concentrations

333

are elevated when residence times over land are low and cannot be explained by primary

334

PCB emissions.

335

PCB re-emissions from ocean water are not a suitable explanation of this finding because

336

this explanation is inconsistent with the opposite seasonal trends reported for PCB-28 at

337

Andøya, another Arctic measurement station that should be equally influenced by PCB

338

re-emissions from ocean waters as Zeppelin. Furthermore the sharp decline in the PCB

339

concentrations observed after a PCB clean-up campaign in 2009 suggests local PCB sources 19 ACS Paragon Plus Environment


340

rather than general emission patterns from water or ice, which are unlikely to show such

341

abrupt changes. The seasonality of the PCB-28 concentrations at Zeppelin is consistent with

342

PCB emissions from local sources, in combination with a seasonal modulation of the PCB

343

emissions by the temperature. Therefore, linking seasonal trends of PCB-28 concentrations

344

from Zeppelin, especially before the cleanup campaign in 2009, to atmospheric long-range

345

transport and emission inventories may be compromised by local PCB sources masking

346

atmospheric transport patterns.

347

The Zeppelin mountain observatory is an important measurement site because it is con-

348

sidered representative of atmospheric background concentrations of various chemicals in the

349

Arctic. According to our findings this representativeness is questionable for PCB-28 con-

350

centrations measured at Zeppelin. The influence of local emissions from Spitsbergen on the

351

concentrations of PCB-28 measured at Zeppelin mountain should be further clarified and

352

taken into account in any future interpretation of the PCB-28 measurements.

353

Acknowledgement

354

We thank Ralf Kägi for helpful comments. Martin Scheringer acknowledges financial support

355

by the Czech Ministry of Education, Youth and Sports (LM2015051) and Masaryk University

356

(CETOCOEN PLUS project, CZ.02.1.01/0.0/0.0/15_003/0000469).

357

References

358

359

360

361

362

(1) UNEP, Final act of the conference of plenipotentaries on the Stockholm Convention on Persistent Organic Pollutants. 2001. (2) AMAP, AMAP Assesssment 2002: Persistent Organic Pollutants (POPs) in the Arctic. Arctic Monitoring and Assesment Programme; 2004. (3) Breivik, K.; Sweetman, A.; Pacyna, J.; Jones, K. Towards a global historical emission 20 ACS Paragon Plus Environment

Page 20 of 26

Page 21 of 26


363

inventory for selected PCB congeners - a mass balance approach 1. Global production

364

and consumption. Environ. Sci. Technol. 2002, 290, 181–198.

365

(4) Kallenborn, R.; Ottesen, R.; Gabrielsen, G.; Schrum, C.; Evenset, A.; Ruus, A.;

366

Benjaminsen, H.; Sagerup, K.; Christensen, G.; Eggen, O.; Carlsson, P.; Johansson-

367

Karsson, E.; Polder, A.; Pedersen, Q., H.R. Lundkvist PCB on Svalbard ; 2011.

368

(5) Eckhardt, S.; Breivik, K.; Manø, S.; Stohl, A. Record high peaks in PCB concentrations

369

in the Arctic atmosphere due to long-range transport of biomass burning emissions.

370

Atmos. Chem. Phys. 2007, 7, 4527–4536.

371

(6) Ubl, S.; Scheringer, M.; Stohl, A.; Burkhart, J. F.; Hungerbuhler, K. Primary source

372

regions of polychlorinated biphenyls (PCBs) measured in the Arctic. Atmos. Environ.

373

2012, 62, 391–399.

374

(7) Hung, H.; Kallenborn, R.; Breivik, K.; Su, Y.; Brorstrom-Lunden, E.; Olafsdottir, K.;

375

Thorlacius, J. M.; Leppanen, S.; Bossi, R.; Skov, H.; Manø, S.; Patton, G. W.; Stern, G.;

376

Sverko, E.; Fellin, P. Atmospheric monitoring of organic pollutants in the Arctic under

377

the Arctic Monitoring and Assessment Programme (AMAP): 1993-2006. Sci. Total

378

Environ. 2010, 408, 2854–2873.

379

(8) Wöhrnschimmel, H.; Scheringer, M.; Bogdal, C.; Hung, H.; Salamova, A.; Venier, M.;

380

Katsoyiannis, A.; Hites, R. A.; Hungerbuhler, K.; Fiedler, H. Ten years after entry

381

into force of the Stockholm Convention: What do air monitoring data tell about its

382

effectiveness? Environ. Pollut. 2016, 217, 149–158.

383

(9) Hung, H.; Blanchard, P.; Halsall, C.; Bidleman, T.; Stern, G.; Fellin, P.; Muir, D.;

384

Barrie, L.; Jantunen, L.; Helm, P.; Ma, J.; Konoplev, A. Temporal and spatial variabil-

385

ities of atmospheric polychlorinated biphenyls (PCBs), organochlorine (OC) pesticides

386

and polycyclic aromatic hydrocarbons (PAHs) in the Canadian Arctic: Results from a

387

decade of monitoring. Sci. Total Environ. 2005, 342, 119–144. 21 ACS Paragon Plus Environment


388

(10) Bossi, R.; Skov, H.; Vorkamp, K.; Christensen, J.; Rastogi, S. C.; Egelov, A.; Pe-

389

tersen, D. Atmospheric concentrations of organochlorine pesticides, polybrominated

390

diphenyl ethers and polychloronaphthalenes in Nuuk, South-West Greenland. Atmos.

391

Environ. 2008, 42, 7293–7303.

392

(11) Eckhardt, S.; Breivik, K.; Li, Y. F.; Manø, S.; Stohl, A. Source regions of some persistent

393

organic pollutants measured in the atmosphere at Birkenes, Norway. Atmos. Chem.

394

Phys. 2009, 9, 6597–6610.

395

(12) Stohl, A.; Hittenberger, M.; Wotawa, G. Validation of the Lagrangian particle disper-

396

sion model FLEXPART against large-scale tracer experiment data. Atmos. Environ.

397

1998, 32, 4245–4264.

398

(13) Stohl, A.; Forster, C.; Frank, A.; Seibert, P.; Wotawa, G. Technical note: The La-

399

grangian particle dispersion model FLEXPART version 6.2. Atmos. Chem. Phys. 2005,

400

5, 2461–2474.

401

402

(14) Stohl, A. Characteristics of atmospheric transport into the Arctic troposphere. J. Geophys. Res. Atmos. 2006, 111 .

403

(15) Hirdman, D.; Sodemann, H.; Eckhardt, S.; Burkhart, J. F.; Jefferson, A.; Mefford, T.;

404

Quinn, P. K.; Sharma, S.; Strom, J.; Stohl, A. Source identification of short-lived air

405

pollutants in the Arctic using statistical analysis of measurement data and particle

406

dispersion model output. Atmos. Chem. Phys. 2010, 10, 669–693.

407

408

409

410

(16) Anderson, P.; Hites, R. OH radical reactions: The major removal pathway for polychlorinated biphenyls from the atmosphere. Environ. Sci. Technol. 1996, 30, 1756–1763. (17) Seibert, P.; Frank, A. Source-receptor matrix calculation with a Lagrangian particle dispersion model in backward mode. Atmos. Chem. Phys. 2004, 4, 51–63.


Page 22 of 26

Page 23 of 26

411

412


(18) Meyer, T.; Wania, F.; Breivik, K. Illustrating sensitivity and uncertainty in environmental fate models using partitioning maps. Environ. Sci. Technol. 2005, 39, 3186–3196.

413

(19) Li, N.; Wania, F.; Lei, Y.; Daly, G. A comprehensive and critical compilation, eval-

414

uation, and selection of physical-chemical property data for selected polychlorinated

415

biphenyls. J. Phys. Chem. Ref. Data 2003, 32, 1545–1590.

416

(20) Kwok, E.; Atkinson, R.; Arey, J. Rate constants for the gas-phase reactions of the OH

417

radical with dichlorobiphenyls, 1-chlorordibenzo-p-dioxin, 1,2-dimethoxybenzene, and

418

diphenyl ether - estimation of OH radical reaction-rate constants for PCBS, PCDDS,

419

and PCDFS. Environ. Sci. Technol. 1995, 29, 1591–1598.

420

(21) Sturm, P.; Tuzson, B.; Henne, S.; Emmenegger, L. Tracking isotopic signatures of CO2

421

at the high altitude site Jungfraujoch with laser spectroscopy: analytical improvements

422

and representative results. Atmos. Meas. Tech. 2013, 6, 1659–1671.

423

424

(22) Kaufman, L.; Rousseeuw, P. J. Finding Groups in Data – an Introduction to ClusterAnalysis; Wiley: Hoboken, 2005.

425

(23) Zhao, Y.; Huang, T.; Wang, L.; Gao, H.; Ma, J. Step changes in persistent organic

426

pollutants over the Arctic and their implications. Atmos. Chem. Phys. 2015, 15, 3479–

427

3495.

428

(24) Klonecki, A.; Hess, P.; Emmons, L.; Smith, L.; Orlando, J.; Blake, D. Seasonal changes

429

in the transport of pollutants into the Arctic troposphere-model study. J. Geophys. Res.

430

Atmos. 2003, 108 .

431

(25) Breivik, K.; Sweetman, A.; Pacyna, J. M.; Jones, K. C. Towards a global historical

432

emission inventory for selected PCB congeners - A mass balance approach-3. An update.

433

Science of the Total Environment 2007, 377, 296–307.



434

(26) Breivik, K.; Armitage, J. M.; Wania, F.; Sweetman, A. J.; Jones, K. C. Tracking the

435

Global distribution of Persistent Organic Pollutants Accounting for E-Waste Exports

436

to Developing regions. Environ. Sci. Technol. 2016, 50, 798–805.

437

(27) Ma, J.; Hung, H.; Tian, C.; Kallenborn, R. Revolatilization of persistent organic pollu-

438

tants in the Arctic induced by climate change. Nature Clim. Change 2011, 1, 255–260.

439

(28) Lohmann, R.; Klanova, J.; Kukucka, P.; Yonis, S.; Bollinger, K. PCBs and OCPs on a

440

East-to-West Transect: The Importance of Major Currents and Net Volatilization for

441

PCBs in the Atlantic Ocean. Environ. Sci. Technol. 2012, 46, 10471–10479.

442

(29) Gioia, R.; Nizzetto, L.; Lohmann, R.; Dachs, J.; Temme, C.; Jones, K. C. Polychlo-

443

rinated Biphenyls (PCBs) in air and seawater of the Atlantic Ocean: Sources, trends

444

and processes. Environ. Sci. Technol. 2008, 42, 1416–1422.

445

(30) Gioia, R.; Lohmann, R.; Dachs, J.; Temme, C.; Lakaschus, S.; Schulz-Bull, D.; Hand, I.;

446

Jones, K. C. Polychlorinated biphenyls in air and water of the North Atlantic and Arctic

447

Ocean. J. Geophys. Res. Atmos. 2008, 113 .

448

(31) Gioia, R.; J., D.; Nizzetto, L.; Lohmann, R.; Jones, K. C. In Occurrence, Fate and

449

Impact of Atmospheric Pollutants on Environmental and Human Health; McConnell, L.,

450

Dachs, J., Hapeman, C. J., Eds.; ACS Symposium Series, 2013; Vol. 1149; pp 3–18.

451

(32) Friedman, C. L.; Selin, N. E. PCBs in the Arctic atmosphere: determining important

452

driving forces using a global atmospheric transport model. Atmos. Chem. Phys. 2016,

453

16, 3433–3448.

454

(33) Herbert, B.; Halsall, C.; Villa, S.; Jones, K.; Kallenborn, R. Rapid changes in PCB and

455

OC pesticide concentrations in Arctic snow. Environ. Sci. Technol. 2005, 39, 2998–

456

3005.


Page 24 of 26

Page 25 of 26


457

(34) Burniston, D. A.; Strachan, W. J. M.; Hoff, J.; Wania, F. Changes in surface area

458

and concentrations of semivolatile organic contaminants in aging snow. Environ. Sci.

459

Technol. 2007, 41, 4932–4937.

460

(35) Jartun, M.; Ottesen, R. T.; Volden, T.; Lundkvist, Q. Local Sources of Polychlorinated

461

Biphenyls (PCB) in Russian and Norwegian Settlements on Spitsbergen Island, Norway.

462

J. Toxicol. Environ. Health A 2009, 72, 284–294.

463

(36) Di Liberto, L.; Angelini, F.; Pietroni, I.; Cairo, F.; Di Donfrancesco, G.; Viola, A.;

464

Argentini, S.; Fierli, F.; Gobbi, G.; Maturilli, M.; Neuber, R.; Snels, M. Estimate of

465

the Arctic Convective Boundary Layer Height from Lidar Observations: A Case Study.

466

Adv. Meteorol. 2012,

467

(37) Tunved, P.; Strom, J.; Krejci, R. Arctic aerosol life cycle: linking aerosol size distri-

468

butions observed between 2000 and 2010 with air mass transport and precipitation at

469

Zeppelin station, Ny-Alesund, Svalbard. Atmos. Chem. Phys. 2013, 13, 3643–3660.

470

(38) Green, N. W.; Skogen, M.; Aas, W.; Iosjpe, M.; Mage, A.; Breivik, K.; Yakushev, E.;

471

Hogasen, T.; Ekhardt, S.; Ledang, A. B.; Jaccard, P.; Staalstrom, A.; Isachseen, P. E.;

472

Frantzen, S. Overvaking av tilførsler og miljøtilstand i Barentshavet og a Lofotenom-

473

radet; 2013.



474

Graphical TOC Entry

475

Zeppelin

Spitsbergen

Mainland (Europa / Asia)


Page 26 of 26

Relationships between atmospheric transport ... - ACS Publications

Recommend Documents