Mercury and metals in South African precipitation

Atmospheric Environment 79 (2013) 286e298

Contents lists available at SciVerse ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Mercury and metals in South African precipitation Susan W. Gichuki a, b, Robert P. Mason a, b, * a b

Department of Marine Sciences, University of Connecticut, Groton, CT 06340, USA Department of Chemistry, University of Connecticut, Storrs, CT 06268, USA

h i g h l i g h t s
Mercury in precipitation in South Africa is comparable to Northern Hemispheric sites.
Metal concentrations are higher at the urban Pretoria site compared to Cape Point.
Relative metal concentrations reflect that of regional particulate.
Concentrations indicate the relative importance of crustal and anthropogenic inputs.
The Cape Point site is impacted by both marine air and terrestrial sources.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 7 December 2012 Received in revised form 27 March 2013 Accepted 5 April 2013

Even though mercury (Hg) is a global pollutant there are few studies of its concentration in the Southern Hemisphere, either in the atmosphere or in atmospheric deposition, and this is particularly true for Africa and developing nations such as South Africa. Emission inventories suggest that there is the potential for anthropogenic impact through elevated deposition in South Africa to sensitive ecosystems. To begin examining such impact, measurements of Hg and other trace metals (Al, Fe, Mn, Co, Ni, Cu, Zn, Cd and Pb) were made of rain collected using a bulk deposition collector at Cape Point, at the tip of Africa, and in Pretoria, a city within the industrial and mining heartland of South Africa. As expected, concentrations and fluxes were elevated in Pretoria, the more urban location. However, there is also evidence that Cape Point site can be impacted by regional pollution during the winter. The volume weighted mean Hg concentration at Cape Point was 10.6 ng L1 compared to 15.8 ng L1 in Pretoria. Comparison of rain concentrations for Hg and other metals, and relative fluxes (normalized to Al), indicate the importance of regional sources of contamination to both sites. The importance of impacted air masses at Cape Point was further investigated using ancillary data (CO and Rn) and back trajectory analysis. Overall, these results suggest that Hg and trace metal concentrations and fluxes are comparable to other locations in the world that are impacted by regional anthropogenic sources. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: Mercury Trace metals Precipitation South Africa Deposition Metal pollution

1. Introduction Even though mercury (Hg) is a global pollutant there are few studies of its concentration in the Southern Hemisphere, either in the atmosphere (Slemr et al., 2011) or in atmospheric deposition, or of its emissions in developing countries, especially in Africa. The same is true for other pollutant metals. As South Africa is the largest producer of primary metals in the world, ranking first in the production of gold (Au), lead (Pb) and copper (Cu) (Masekoameng

* Corresponding author. Department of Marine Sciences, University of Connecticut, Groton, CT 06340, USA. Tel.: þ1 860 4059129. E-mail address: [email protected] (R.P. Mason). 1352-2310/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.atmosenv.2013.04.009

et al., 2010), there is substantial potential for these emissions from human activity to contribute to local and regional contamination. Overall, 40% of the world’s gold reserves are in South Africa and the country is responsible for 12% of global production of this precious metal. The country also relies on coal combustion for much of the electricity production (64% of the energy supply; Dabrowski et al., 2008), and for industry, and this could contribute substantially to Hg emissions. Mercury, which is emitted as both elemental mercury (Hg0) and ionic mercury (HgII) (Sprovieri et al., 2010), and other metal emissions from these activities could be substantial. There are two fractions of atmospheric HgII operationally defined according to their physicochemical properties and current measurement techniques; the gaseous ionic HgII fraction termed reactive gaseous mercury (RGHg) and particulate

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

associated mercury (HgP) (Landis et al., 2002). The speciation of RGHg is thought to consist of ionic neutral complexes resulting from reactions with halides and other oxidants (Holmes et al., 2009; Mason and Sheu, 2002). There has been growing concern over the increasing emission levels of mercury in the world. One global anthropogenic Hg emissions inventory for the year 2000 (Pacyna et al., 2006) ranked China as the number one Hg emitter in the world and South Africa as second, with South African emissions being attributed to gold mining and stationary sources such as coal combustion. However, recent evaluations (Dabrowski et al., 2008; Leaner et al., 2009) suggest that previous estimates were high by more than an order of magnitude, and that emissions from coal fired power plants in South Africa were 9.8 tons per annum, and approximately 50% of the total emissions. Other estimates for South Africa (Pirrone et al., 2010; Streets et al., 2011) are also lower than the initial Pacyna et al. (2006) value, but suggest emissions from Africa are increasing. Gold mining operations in South Africa mainly use the MacArthur Forrest process that uses cyanide instead of mercury amalgamation for extraction (Naicker et al., 2003). However, there are traces of Hg in the ore that could be released during processing but the overall emission is considered small (Leaner et al., 2009). While many developed countries have established monitoring programs, little has been done in developing countries especially in Africa. Measuring deposition is one way of evaluating the impact of Hg emissions and while both wet and dry deposition of mercury is important, it has been shown that quantifying the amount of mercury in precipitation is an effective way of estimating the relative fluxes (Schmeltz et al., 2011), and comparing across global locations. In addition, properly chosen monitoring stations can provide more accurate estimates of wet deposition at regional scales that can help constrain atmospheric mercury fate and transport models (Zhang et al., 2009). Recent modeling of emissions and deposition suggest that net Hg deposition is not substantially higher around South Africa than over other continents at comparable latitudes, or in much of the Northern Hemisphere remote from major anthropogenic inputs (Corbitt et al., 2011). The data presented here provides a snapshot for the interior (Pretoria) and a coastal region (Cape Point) of South

287

Africa for Hg. Measurements of atmospheric total gaseous Hg at Cape Point suggest that concentrations are somewhat lower than the Northern Hemisphere, varying between 0.8 and 1.0 ng m3 (Slemr et al., 2011), with an indication of decreasing concentrations since the early 1990’s. This could suggest, contrary to the models (e.g. Soerensen et al., 2010), that Hg deposition may be lower at Cape Point than at locations in the Northern Hemisphere. The precipitation measurements allowed an investigation of these ideas and the comparison of the relative ratios of metals in precipitation allowed an examination of the potential sources of Hg to the sampling sites, and an examination of South Africa deposition in a global context. 2. Materials and methods 2.1. Study area, sampling, and analysis protocol Rain samples were collected from two sites: Cape Point and Pretoria (Fig. 1). The Cape Point site is located at the southern tip of the Cape Peninsula within the Cape of Good Hope nature reserve atop a peak 230 m above sea level and about 60 km south of Cape Town (34 210 S, 18 290 E). It also functions as a Global Atmospheric Watch (GAW) site and is operated by the World Meteorological Organization (WMO) (Slemr et al., 2008). The site is thought to receive clean air masses most of the time and was chosen to provide a Southern Hemisphere background signal. The site in Pretoria (25 440 S, 28 160 E) was located on a building rooftop at the Council for Scientific and Industrial Research (CSIR) campus. Pretoria is an urban area located 50 km north of Johannesburg, the commercial and industrial hub of the country. Much of the gold mining and other precious metal mining occurs in the vicinity of Johannesburg. Pretoria is also in relative close proximity to industrial activity in the northern reaches of the country where many of the coal-fired power plants are located in proximity to the major coal mining areas. Additionally, steel and other metal refining is also concentrated northwest of Pretoria (Masekoameng et al., 2010). Precipitation samples were collected between June 2007 and December 2009. The sampling setup comprised a homemade glass

Fig. 1. Maps showing sampling sites in South Africa (blue markers). In the left Panel: Pretoria and Cape Point, in the right panel: the site at Cape Point shown in relation to Cape Town. Maps taken from Googlemaps copyright 2012 and reprinted with permission from Google.

288

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

funnel connected to a PTFE (Teflon) bottle. Sampling was done weekly and the equipment was transferred between the two sites because the rain falls in alternate seasons i.e. in the Cape region, the rain falls mostly in the winter (Mediterranean climate), whereas the winters in Pretoria are dry. Conversely, in the summer, rain falls in Pretoria, whereas the Cape region is dry. This strategy collected most of the rainfall as there is very little rainfall in summer on the Cape Peninsula (<2 cm month1 NoveFeb; March w2 cm month1) and little precipitation in winter in the region of Pretoria (<2 cm month1 on average MayeAugust) so we missed <20% of the precipitation at both locations. Each sample was preserved after collection by spiking with 0.5% by volume trace metal grade hydrochloric acid (HCl) prior to refrigeration and shipping to the University of Connecticut, Avery Point campus for analysis. On arrival at Avery Point, the bottles were weighed so that the rain weight (volume) could be determined by subtracting the bottle weight. To prepare samples for analysis, 1 mL of 0.5% (90 mM) bromine monochloride (BrCl) (v/v) was added irrespective of sample volume, which depended on the amount of rain per week. The oxidation/digestion step proceeded overnight. Afterward the sample was pre-reduced with hydroxylamine hydrochloride (NH2OH$HCl) to destroy the excess bromine, prior to ionic Hg(II) being reduced to Hg0 with stannous chloride (SnCl2) (US EPA Method 1631, Lamborg et al., 2012) and sparged from solution and trapped on a gold-coated sample trap. A TekranÔ Cold Vapor Atomic Fluorescence Spectrometer (CVAFS) was used for sample analysis. The detection limit (DL) for rain was calculated as three times the standard deviation of the blank values. The DL for Cape Point was 1.5 ng L1, whereas that of Pretoria samples was 1.2 ng L1. The overall detection limit for all samples was 1.4 ng L1. The average relative percent difference (RPD) for duplicate runs of samples was 2.8% for Cape Point, and 7.7% for Pretoria. In addition to THg analysis, trace metal analysis was completed whenever sufficient sample volumes were available. Calibration standard solutions were prepared from a multi element metals solution purchased from SPEX Certiprep. Metal analysis of acidified samples was done on a Perkin Elmer Elan DRC II Inductively Coupled Plasma-Mass Spectrometer (ICP-MS) coupled to an autosampler. The rain samples were introduced to the plasma by means of a peristaltic pump, via a nebulizer and spray chamber. The instrument was optimized daily according to manufacturer’s instructions to ensure proper functioning at optimal conditions. Optimization involved plasma alignment, mass calibration and auto-tuning using a multi-element tuning solution (Ba, Be, Ce, Co, In, Pb, Mg, Tl, Th, all at 1 ug L1 in 5% (v/v) nitric acid). The acceptable performance criteria of the ICP-MS are: CeO/Ce < 3%, Ba2þ/Ba < 3%; sensitivity of the 1 mg L1 Indium solution >40,000 counts per second (cps); and the instrument background level of m/ z 220 < 2 cps. The detection limits for each metal analyzed were as follows: (Al: 0.70; Co: 0.31; Cu: 0.69; Mn: 1.83; Zn: 2.58; Ni: 0.31; Fe: 4.55; Cd: 0.02; Pb: 0.10; all values in ng L1). The operational criteria used for the ICP-MS analysis were as described in the operational manual, with some minor modifications depending on the method being used for analysis (see Supporting Information (SI) Table S1). 3. Results 3.1. Total Hg (THg) in precipitation and deposition fluxes Fig. 2 and Table S2 in the Supporting Information (SI) summarize the concentrations of Hg in rain and fluxes at the two sites in South Africa and show that the average concentration in Pretoria is

Fig. 2. A comparison of the volume weighted mean concentration (VWM) (ng L1) of Hg in rain at Cape Point and Pretoria from 2007 to 2009, the average concentration (ng L1) and the average weekly flux in ng m2 wk1. The error bars are based on the standard error of the mean.

approximately 150% of that at Cape Point, although the differences in concentration are not statistically significant given the large standard deviations (Table 1). The difference is larger for the flux illustrating the importance of rainfall amount in determining Hg deposition flux. The differences in the flux are statistically significant, using the same statistical analysis as used for Table 1. This is further illustrated in the plots comparing the results at both locations in Fig. 2. It should be noted that these concentrations and fluxes represent total or ‘bulk’ deposition since the rain funnel was open to the atmosphere at all times. This means that particulates and RGHg from the atmosphere could have dry deposited on the funnel surface in between rainfall, contributing dry deposition that would then be washed into the collection bottle by rain. According to estimates by Lamborg et al. (1999), the particulate dry depositional flux is 0.4e1.0% of the overall flux from the non-urban atmosphere so it is unlikely that any particulate deposition would substantially alter the flux except in locations adjacent to urban environments (Mason et al., 2002). However, some estimates for coastal environments predict that dry deposition of RGHg could be equivalent to that of wet deposition (e.g. Holmes et al., 2009; Laurier and Mason, 2007), and if so this could substantially alter the flux. It is not clear what fraction of any RGHg deposited to the funnel could be reduced and reemitted but this is likely a small fraction. Therefore, actual wet deposition may be lower than the total deposition fluxes calculated here. However, the Soerensen et al. (2010) GEOS-Chem model predicts the wet deposition of Hg at Cape Point as 24 nmol m2 yr1, which is equivalent to 185 ng m2 wk1, assuming rainfall over 26 weeks, on average, based on historical statistics. This is similar to the estimate in Fig. 1 suggesting that dry deposition does not contribute substantially to the flux measured using the open collector. Also, the model estimates for dry particulate and RGHg deposition are about 25% of the wet flux (Soerensen, pers. comm.) again suggesting that dry deposition would not contribute substantially to the fluxes obtained with the open collector. Similarly, a study in Florida where open collectors were co-deployed with automatic collectors also found little statistical difference in the bulk versus wet collector Hg fluxes (Guentzel et al., 1995, 2001) and others have found similar results (Iverfeldt and Munthe, 1993). Given these previous results and the comparison between the measurements and models, we conclude that the measurements using the open collectors primarily reflect the wet deposition Hg flux at this location.

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

289

Table 1 Mean trace metal concentrations at Cape Point and Pretoria (mg L1) as volume weighted mean (VWM) values, and as an average and standard deviation, and the relative differences in the averages and their significance. Metal

VWM (mg L1) Pretoria

Av (mg L1) Pretoria

St. Dev (mg L1) Pretoria

VWM (mg L1) Cape Town

Av (mg L1) Cape Point

St. Dev (mg L1) Cape Point

Rel % diff.*

Signif.** 95% CI

Al Fe Mn Co Ni Cu Zn Cd Pb Hg

58.3 45.0 9.9 0.22 0.58 1.9 10.3 0.03 1.2 0.016

78.0 62.3 13.7 0.30 0.82 2.5 11.1 0.04 1.5 0.019

66.0 63.3 14.5 0.27 0.94 2.4 13.0 0.04 1.5 0.009

9.0 44.9 2.0 0.19 8.9 0.81 56.6 0.02 1.0 0.011

9.9 31.8 1.8 0.17 8.3 0.81 68.5 0.02 1.2 0.011

7.6 77.5 2.7 0.36 19.2 0.55 86.3 0.02 2.2 0.014

550 <1 410 16 94 130 80 50 16 50

Yes No No No No Yes No No No No

Notes: * Rel. % Diff. ¼ 100(CPretoria  CCapePoint)/CCapePoint. ** Signif. ¼ Tukey’s test of significant difference (two sided test with a 95% confidence interval).

While there have been other measurements of Hg made in the aquatic environment and in air in South Africa (Walters et al., 2011; Slemr et al., 2008; Papu-Zamxaka et al., 2009; Kading et al., 2009), these are the first measurements of Hg in rain in South Africa. Fig. 3 shows the weekly volume-weighted concentrations and deposition fluxes at Pretoria, while Fig. 4 shows comparable data for Cape Point. There are no obvious trends over time in terms of changes in concentration and this is consistent with the findings by other investigators. If “wash-out” of particulate or RGHg is the major contributor to the Hg in rain at these locations, then there could be a “scavenging effect” where low rainfall events have higher concentrations due

to the removal of the particulate and RGHg in the initial rainfall. This has been shown in other studies (Guo et al., 2008; Hall et al., 2005; Mason et al., 2000). It has also been shown that during rainfall RGHg concentrations are very low (e.g. Laurier and Mason, 2007) suggesting that this fraction is readily scavenged. Under such scenarios, during large rain events, the THg wet deposition flux increases although continuous rain dilutes the THg concentration (Seo et al., 2012). The relationships between rainfall amount and concentration for both sites are shown in Fig. 5. It should be reiterated that as the rain collections are weekly, there is the potential for more than one rain event per week, and this could obscure any trends that exist. However, even so, higher

Fig. 3. a) Weekly concentrations of total mercury (THg) (ng L1) in rain at Pretoria. Missing values indicate that there was no rain or a sample was not collected; b) Total deposition flux by week (ng m2 wk1) for each week.

290

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

Fig. 4. a) Weekly concentrations of total mercury (THg) (ng L1) in rain at Cape Point. Missing values indicate that there was no rain or a sample was not collected; b) Total deposition flux by week (ng m2 wk1) for each week.

amounts of rainfall are associated with lower THg concentrations overall. If there are continual sources of Hg to the air mass during a rain event, or if the scavenging does not completely occur, then there

should be a relationship between the Hg flux and the amount of rainfall. Figure S1 in the SI shows the relationship between rainfall depth and deposition flux at both locations. Linear regression analysis of this data yielded an R-value of 0.597 (p < 0.05) for

Fig. 5. Rainfall depth (cm) versus THg concentration (ng L1) for Pretoria and Cape Point respectively.

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

Pretoria and 0.428 (p < 0.05) for Cape Point, demonstrating significant positive relationships between rainfall amount and total deposition flux at both locations. However, there are clearly other factors that influence the overall flux of Hg, such as the impact of anthropogenic and other sources. 3.2. Trace metals in precipitation Trace metal concentrations in the atmosphere can be good indicators of pollution from anthropogenic activities as they are often associated primarily with a particular source category/activity (e.g. Se emissions are primarily from coal). However, although human activities, such as fossil fuel combustion, release metals such as Cu, Zn and Cd into the atmosphere, other metals, such as Al, Fe and Mn, do not have a dominant anthropogenic source, and most of the aerosol in the atmosphere comes from natural sources (soil and bedrock particulate) where they are naturally abundant (Vuai and Tokuyama, 2011). By looking for relationships or correlations between these metals and with Hg, and by examining ratios to socalled “crustal” elements (Al, Fe), we can identify the importance of contamination sources in contributing to the metals collected in rainfall over time. Table 1 summarizes the concentration data for the metals while the individual data are contained in Tables S3 and S4 of the SI for Pretoria and Cape Point respectively. To put the values in context, the relative concentrations of the metals to Al are compiled in Table 2, and these values compared with those of some potential natural sources (crustal particulate and sea salt) as well as data collected by others for atmospheric particulates in locations in South Africa (Sekonya, 2009). These locations are Khayelitsha, a rapidly expanding urban environment with both formal houses and so-called “informal” dwellings built from a variety of materials on the outskirts of Cape Town, about 25 miles northwest of Cape Point, and Ferrobank, a location within the industrial heart of the northwest region of South Africa, about 65 miles west of Pretoria, where iron smelting and refining and other industrial activities using coal are concentrated. The relative difference in the concentrations of the metals provides some initial indication of potential sources and the

Table 2 Average values (plus standard deviation) for all the data for each location calculated from the ratios of each element measured to that of aluminum in each weekly sample. Also shown are the range in values for these metal in crustal rock and in sea salt, compiled from the literature (Chester, 2003 and references therein). Also shown are the ratios found for atmospheric aerosols collected at Khayelitsha and at Ferrobank, two locations proximate to the rain sampling sites, taken from Sekonya (2009). Element (E)

(Av E/Al) precipitation

St Dev (E/Al) precipitation

(E/Al) sea salt

(E/Al) Khayelitsha

Cape Pt. Co Cu Mn Zn Ni Fe Cd Pb

0.019 0.15 0.20 24.9 0.87 3.80 0.0041 0.21

0.033 0.22 0.25 60.3 1.70 9.79 0.0067 0.39

0.5e4 0.01e0.1 0.1e0.4 0.1e5 0.06 1e3.4 6e8  103 0.01e1

e 121 34.1 91.8

Pretoria Co Cu Mn Zn Ni Fe Cd Pb

3.8  103 0.037 0.16 0.18 0.010 0.76 8.2  104 0.022

1.3  103 0.022 0.055 0.21 3.5  103 0.28 1.1  103 9.7  103

2.46  104 6.77  104 0.012 8.61  104 9.84  104 0.62 1.35  106 1.6  104

628 36.5

0.033 0.28 0.22 2.9 0.11

291

validity of the results. For most of the metals the concentrations are within a factor of two between sites (rel. diff.  100%; Table 1) suggesting similar sources, either crustal or anthropogenic. However, statistically significant differences in concentration between the two locations were only found for Al and Cu. For some metals such as Zn and Ni, the mean concentrations are substantially different but the standard deviations are also large. The lack of difference is further confirmed by the comparisons in Table 2 which shows the ratio of each metal (E) in each sample relative to that of Al (E/Al). These indicate how well the relative rain concentrations correspond to those of regional aerosols, which should be the primary source of the metal in deposition. For Al, the concentration is much higher in Pretoria and this likely reflects a higher dust component in this more arid location. Similarly, Mn is also elevated in Pretoria but this is not the case for Fe. Contrary to the other metals, the concentrations of Zn and Ni are higher at Cape Point. An examination of the individual data (Table S4) shows that there are high concentrations of both elements on occasion. It is possible that these elevated concentrations represent local contamination. Potential contaminants sources could be the GAW sampling equipment as the air sampling tower is galvanized or from nearby buildings and human activity, or the local area, which is a tourist attraction although access to the sampling site is limited (e.g. diesel-derived Ni). If these higher values are not considered (>20 mg L1 Zn; >5 mg L1 Ni), then there is little difference between the two locations in the concentration of these metals. A comparison of the metal ratios to Al (Table 2) in precipitation in Pretoria to the aerosols from Ferrobank show comparable values in most cases, except for Fe. This is not the case for Cape Point as the ratios are much higher for Khayelitsha aerosols, which are also much higher than those at Ferrobank. The Khayelitsha aerosols were low in Al. This indicates that the local contamination sources (wood and coal heating/cooking, vehicle emissions) at this location overwhelm the regional aerosol signal. The ratios for most metals at Cape Point are more similar to those at Ferrobank, likely indicating the importance of regional aerosols in contributing to precipitation even at Cape Point, even though it is considered to be remote from pollution sources.

4. Discussion 4.1. Comparison of precipitation results with other locations around the world To put the South African data in context, comparison is made to other locations around the world in Table 3. Measurements have been made in some other sites in the Southern Hemisphere, most notably in Australia (Dutt et al., 2009) and Brazil (Lacerda et al., 2002). Lacerda et al. (2002) attribute the low values at the Brazil sites to the lack of point sources and the strong influence of marine air masses along the Rio de Janeiro State coastline. The site at Mace Head, another WMO site located to sample primarily marine air, is most similar to Cape Point but in the Northern Hemisphere, as it is located on the west coast of Ireland, 88 km away from Galway city. It has a clean sector zone from 180 to 3000 with open access to the North Atlantic Ocean, which represents marine background conditions for atmospheric mercury and other background trace gases (Ebinghaus et al., 1999). When compared to U.S. east coast data from the Mercury Deposition Network (MDN) (http://nadp.sws. uiuc.edu/mdn/), concentrations at Cape Point are comparable (Table 3) while concentrations in Florida are similar to Pretoria. The same applies to deposition fluxes. The MDN samples are collected on a weekly basis, hence making a direct comparison easier. Also, most MDN sites are located away from direct point source impact.

292

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

Table 3 A comparison of the concentrations of mercury (volume weighted mean (VWM)) for Pretoria and Cape Point with other studies. For deposition flux, data from the literature are converted to similar units (ng m2 wk1). In some instances the mean (average) and standard deviation are given as these were the values published. The referenced literature is indicated below the table. Location Urban NSW, Australia Toronto, Canada Chongqing, China Guizhou, China Remote sites, China Coastal Mace Head, Ireland Brazil Japan, various California, USA Bermuda North America Underhill, VA MDN, site FL11 MDN, site SC05 MDN, site MD99 MDN, site NJ30 MDN, 13 sites in New England New Hampshire, USA This Study Pretoria, South Africa Cape Point, South Africa

Sampling type and period

THg (ng L1) range: VWM or mean  sd

Range or mean flux (ng m2 wk)

Daily, 6/2006e12/2007 Event, 2006e2008 Event, 7/2010e6/2011 Monthly, 2006 Various, 2005e2009

1e19 4.0e122 5.8e158; 30.7 7.5e149; 36 1-31

21e1700 80e2140 640 710

Event, 11e15/9/95 Event, 10/1998e12/1999 Biweekly, 2003e2005 Event, 2007e2008 Weekly, 2008e2009

4e12.2 <0.02e2.95 2e18; 5.8 4.7

140

6 7 9 13 14

Event, 1/93e12/03 Weekly, 2007e2010 Weekly, 2007e2010 Weekly, 2007e2010 Weekly, 2007e2010 Weekly, 1996e2002 Weekly, 7/2006e8/2009

7.8e10.5 14.4  11.3 8.3  5.0 13.0  10.7 8.7  8.4 3.8e9.4* 1.4e65

190 510  690 190  200 220  210 190  210 70e200* 210

10 11 11 11 11 12 8

Weekly, 2007e2009 Weekly, 2007e2009

3.8e60.7; 15.8 1.2e52.5; 10.6

390 170

14 14

8e63 250

Ref

1 2 3 4 5

References: 1: Dutt et al. (2009); 2: Zhang et al. (2012); 3: Wang et al. (2012); 4: Guo et al. (2008); 5: taken from ref. 3; 6: Ebinghaus et al. (1999); 7: Lacerda et al. (2002); 8: Lombard et al. (2011); 9: Sakata et al., 2006; 10: Keeler et al., 2005 11: MDN website; 12: Vanarsdale et al., 2005; 13: Conaway et al., 2010; 14: This study and our unpublished data. Notes: * range in VWM and weekly deposition for each site across years; ** averages for entire sampling period.

Further, rainfall amounts are known to influence deposition fluxes. For instance, in the Australian study (Dutt et al., 2009), the average annual rainfall depths were 94 mm for Sydney (urban) and 60 mm at Hunter Valley (near a coal fired power station). At Pretoria, the long-term average annual rainfall, from weather data (www.tutiempo.net), varies between 40 and 100 cm. Rainfall at Cape Point was 29 cm in 2009. Cape Town rainfall varies from 40 to 70 cm yr1 (long-term average 51 cm yr1) but Cape Point receives less rain than Cape Town and the rest of the peninsula. Rainfall differences are one reason why sites with similar THg concentrations in rain have different deposition fluxes. The flux measured at Cape Point is similar to that of remote sites in eastern North America (Table 3) and comparable to those of other coastal locations. Concentrations and fluxes are higher in Florida, USA than the other east coast sites and Cape Point. The data from the Brazilian study appear to be anomalously low compared to the other data in Table 3. The small differences in the precipitation concentrations for the coastal WMO sites (Cape Point and Mace Head) contrasts the differences in atmospheric total gaseous Hg (1.4e1.6 ng m3 at Mace Head and 0.8e1.0 ng m3 at Cape Point for 2007e2009) (Slemr et al., 2011). This suggests that atmospheric chemistry is more important than absolute concentration in driving the concentrations of Hg in precipitation at these remote coastal sites. However, an alternative explanation is that the Cape Point site is more impacted by local particulate sources, such as the city of Cape Town. The flux for Pretoria is elevated compared to Cape Point but is within the range of the other sites in urban and/or impacted locations (Toronto, the Australian and Asian sites). It is however lower than the two impacted locations in China listed in Table 3. Overall, the range in values in both concentration and flux for the South African sites are comparable to similarly impacted sites in similar locations around the world. The differences in Hg flux between the coastal Cape Point site and the more impacted Pretoria site, a factor of 2e3, is of similar magnitude to that found in a regional study in Maryland, USA

where concentrations and fluxes in Baltimore were twice that found at Solomons, MD, a coastal site about 60 miles south of Baltimore and a similar distance from Washington, DC (Mason et al., 2002). For the other metals, that study found that Pb was nearly three times higher at the urban site while the other metals had relatively similar concentrations at both sites. This contrasts the South African data (Table 1) where concentrations of Ni and Zn were 5e10 times higher at Cape Point than in Pretoria, but Cu and some of the crustal elements were higher in Pretoria. The trace metals (Al and Fe) concentrations for both Pretoria and Cape Point reflect a mixed influence of crustal sources and anthropogenic emissions (Song and Gao, 2009) at both locations. It is likely that local sources are the cause of the elevated Zn and Ni data at Cape Point. The South African results for metals in precipitation are similar to findings in areas in or close to urban environments such as Newark, NJ (Song and Gao, 2009) and Okinawa Island and other Japanese sites (Vuai and Tokuyama, 2011) (Table 4), and coastal locations, such as the Chesapeake Bay, as noted above. However, the Cape Point data are elevated relative to measurements of remote ocean waters suggesting that the site is impacted by local and regional sources and does not reflect an oceanic background site. The concentrations of the rest of the metals are not markedly different from measurements in other parts of the world (Table 4). 4.2. Relationships between trace metals and links to potential contamination sources To examine further the relationships between metals, which may further indicate potential sources, the interactions of these metals were investigated using Pearson correlation coefficients as shown in Tables 5 and 6. At Cape Point, Hg did not correlate with any other metal except for Pb. This is not surprising given that the particulate flux, and the scavenging of particulate by precipitation, is not the dominant source of Hg in rain (Lamborg et al., 1999) as it

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

293

Table 4 Trace metal concentrations at Pretoria and Cape Point compared to other locations around the world at similar locations (coastal or urban). Concentrations are volumeweighted averages (mg L1). For some locations, average values for multiple sites are given rather than volume weighted values. Location

Al

North Atlantic* North Pacific ** South Pacific# Newark, NJ Okinawa Is, Japan Japan, various Paradise, NZ Chesapeake Bay Baltimore Florida sites (5)$ Avery Point, CT Pretoria Cape Point

Co

2.1 16 9.54 2.71

0.02 e

e 16.8 26.9 4e114 6.94 58.3 9.04

e e e 0.02 0.22 0.19

Cu

Mn

Zn

0.66 0.013 0.021 2.82 1.29 1.3  0.9 0.013 0.91 1.22 0.3e1.4 0.40 1.86 0.81

0.27 0.012 0.02 e 2.01 3.1  1.1 0.073 1.57 6.91 1.5e4 0.88 9.94 1.96

1.15 0.052 1.6 6.60 9.25 12 0.038 2.42 6.27 4e9 1.57 10.31 56.6

Ni

0.55 e 0.46  0.15 e 0.58 0.73 1.3e3.2 0.39 0.58 8.94

Fe

Cd

Pb

Ref.

4.8 1.0 0.42 8.35 2.36

0.06 2  104

0.77 0.035 0.014 0.47 e 2.9  1.3 0.02 0.54 3.51 0.3e0.7 0.65 1.16 1.00

1 1 1 2 3 6 4 5

2.1 9.36 20.8 24e80 3.86 45.0 44.9

0.03 e 0.10  0.04 3.6  104 0.04 0.05 0.01e0.08 0.019 0.03 0.02

8 7 7 7

Notes: * Bermuda; ** Enewetak Atoll; # Samoa Island; $ concentrations estimated assuming 1.3 m yr1 rainfall at all sites References: 1: Chester (3003) and references therein; 2: Song and Gao (2009); 3: Vuai and Tokuyama (2011); 4: Halstead et al. (2000); 5: Mason et al. (2002); 6: Sakata and Asakura (2009); 7: This study and unpublished data; 8: Landing et al. (1995).

is for metals in general. The Hg in precipitation at this location is derived from both terrestrial or ocean air masses and is likely produced by photochemical reactions in the troposphere (Holmes et al., 2009). Similarly, no correlation between Hg and other metal concentrations in precipitation was found for a coastal US site (Chesapeake Bay; Mason et al., 2002), and a study in Florida across 5 sites also found little correlation between Hg and other metals (Landing et al., 1995). At Cape Point, Fe, a crustal element, correlates with Cu, Co, and Mn indicating that these metals, while having an anthropogenic source, are also derived from crustal sources. Similarly, in the study around the Chesapeake Bay (Mason et al., 2002) most of the metals (Al, Fe, Mn, Cu, Cd, Zn and Pb) were reasonably well correlated at the more remote site in Solomons, MD. In Florida, the crustal metals (Al, Fe and Mn) were strongly correlated but the heavy metals and metalloids were not all strongly correlated with each other, suggesting a variety of sources for these metals (Landing et al., 1995). The stronger relationship between some metals (Co, Cu, Mn and Ni), indicates a commonality in sources; most likely anthropogenic. High temperature processes such as non-ferrous metal smelting and fossil fuel combustion are known to liberate metals such as Cu, Ni and Pb (Herrera et al., 2009). Some metals (e.g. Cu and Zn) are also associated with wearing of vehicle parts such as brakes and tires (Thorpe and Harrison, 2008), and metals can also be released from oil refining, which occurs within the greater Cape Town region, or in association with vehicle tailpipe emissions. Mixed sources are likely. For example, Mn in atmospheric waters is derived from windblown dust, anthropogenic activities such as waste incineration, iron and steel manufacture, mining and fossil fuel combustion (Willey et al., 2009).

At Pretoria, there are a significant number of strong relationships between metals of crustal origin and anthropogenic derived metals. Most notably, Hg exhibits a strong relationship with Al, Co, Cu, Mn, Ni, Fe, Cd and Pb, which is not the case for Cape Point. This is further evidence that in Pretoria, Hg emissions come from a variety of anthropogenic processes, and potentially that Hg in precipitation is more strongly related to particulate scavenging at this inland location. It is also worth noting that at this location, Pb has strong relationships with Al, Co, Cu, Mn, Ni and Fe. Lead is associated with many industrial processes especially as a result of the introduction of tetraethyl Pb in gasoline in the 1920’s (Cho et al., 2011). Its use has gradually been phased-out worldwide but the complete phaseout of leaded gasoline in South Africa was only achieved in 2006. Therefore, there is a likelihood that Pb is still present in the surface soils, which could be contributing to Pb in precipitation. Also, Pb was replaced with methylcyclopentadienyl manganese tricarbonyl (MMT) in gasoline in 2000 as Pb was phased out. MMT also functions as an octane booster (“anti-knock” additive) and could result in elevated levels of Mn in areas where used (Batterman et al., 2011). It is worth noting that the volume weighted average concentration of Mn is approximately 5 times higher at Pretoria than at Cape Point (Table 1). As discussed above, it is clear from examination of the ratios (Table 2) that there is a strong similarity in the ratios in Pretoria rain and in the particulate collected at Ferrobank for most metals (within a factor of 5). These ratios are also mostly much higher (10e100 times) than that of terrestrial material, even for some of the crustal materials such as Mn. Only Fe has a ratio in Pretoria rain similar to its crustal abundance. Overall, this confirms the stronger anthropogenic influences in Pretoria but suggests that there are anthropogenic sources impacting Cape Point as well.

Table 5 Pearson correlation coefficients for trace metals in precipitation collected at Cape Point. Bold values indicate correlations that are significant for the number of observations.

Hg Al Co Cu Mn Zn Ni Fe Cd Pb

Hg

Al

Co

Cu

Mn

Zn

Ni

Fe

Cd

Pb

1.000 0.068 0.540 0.235 0.514 0.172 0.059 0.272 0.356 0.569

1.000 0.183 0.489 0.376 0.245 0.174 0.045 0.077 0.130

1.000 0.828 0.971 0.047 0.994 0.620 0.143 0.069

1.000 0.857 0.056 0.803 0.381 0.212 0.260

1.000 0.034 0.975 0.530 0.110 0.093

1.000 0.023 0.212 0.201 0.195

1.000 0.587 0.116 0.094

1.000 0.170 0.166

1.000 0.004

1.000

294

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

Table 6 Pearson correlation coefficients for trace metals in precipitation collected in Pretoria. Bold values indicate correlations that are significant for the number of observations.

Hg Al Co Cu Mn Zn Ni Fe Cd Pb

Hg

Al

Co

Cu

Mn

Zn

Ni

Fe

Cd

Pb

1.000 0.613 0.769 0.783 0.808 0.422 0.775 0.556 0.609 0.781

1.000 0.923 0.822 0.857 0.272 0.837 0.911 0.330 0.884

1.000 0.933 0.963 0.285 0.951 0.875 0.465 0.974

1.000 0.938 0.399 0.930 0.834 0.369 0.963

1.000 0.310 0.992 0.760 0.419 0.969

1.000 0.336 0.232 0.261 0.342

1.000 0.728 0.448 0.958

1.000 0.242 0.861

1.000 0.429

1.000

The situation is more complex at Cape Point. Firstly, the variance in the ratio data is much higher for the Cape Point samples (% RSD > 100% for all metals) compared to that of Pretoria (RSD 30e 60% for all metals except Zn and Cd) suggesting that there is less consistency in the sources of the metals to rain, and that the ratios vary widely between weekly collections. This is evident from an examination of the individual data. In terms of sources, comparison of the ratios to Al for the rain compared to the range of values measured in sea salt (Table 5) suggest that for many of the metals regional marine air is an important source, but there is also anthropogenic influence. However, in many cases the values are at the high end of the range of sea salt values, which suggests that these aerosols are impacted by anthropogenic sources. This idea is reinforced by comparison of the ratios to that measured at Khayelitsha, which has a much higher ratio for the metals measured compared to both the Cape Point rain and the sea salt ratios. This suggests that the Cape Point site is not “pristine”. The concentration data suggest that it is not as highly impacted as the Pretoria site, although there are few differences that are statistically significant. The ratios of the average values (E/Al Cape Point/E/Al Pretoria) for each metal are similar (range 1e10) except for Zn and Ni. This analysis confirms the notion expressed above that there are local sources of Zn and Ni that are contributing to these metals in the rain at Cape Point. 5. Linking rainfall concentrations and air mass sources The GAW site at Cape Point was setup in 1977 with the primary role of monitoring long-term trends of trace gases in air representing background conditions, not affected by local or regional pollution. However, although the dominant wind direction is from the southeastern sector, which is representative of clean maritime air from the Southern Ocean, Cape Point is also subjected to air from the north to northeastern sector mainly during the austral winter (Brunke et al., 2010, 2012). This suggests that the site can receive polluted continental air masses, or relatively local sources given its proximity to Cape Town. Overall, most of the polluted air masses originate over the continent (due north) but some originate from the east (Brunke et al., 2010), with the predominance of the pollution events arriving during the day. Most of these events occur between March and August and therefore coincide with the sampling period at Cape Point. In addition to the rain collections done for this project, Hg0 is monitored at the site as well as a variety of trace gases (Brunke et al., 2012). In addition to pollution events (high Rn, CO and Hg0 concentrations), Hg0 depletion events (MDE’s) (i.e. depleted Hg0 due to enhanced local atmospheric chemistry) have been reported (Brunke et al., 2004, 2010; Slemr et al., 2008). Pollution events involve the influx of continental air masses whereas depletion events constitute rapid atmospheric oxidation processes that lead to the production of RGHg and the depletion of Hg0,

which has a high potential for deposition. Most depletion events occur in the afternoon, indicative of the importance of atmospheric chemistry, and occur during winds from a variety of sectors but not from a southerly direction. This indicates, as suggested by others (Laurier and Mason, 2007; Engle et al., 2008), that the formation of RGHg is favored when there is a mixture of oceanic and polluted air masses. The occurrence of MDE’s may explain some of the more elevated concentrations recorded in rain at Cape Point. In addition, it explains why no correlation was found between Hg and other trace metals of anthropogenic origin in Table 5 if a large fraction of the Hg is from in-situ atmospheric production within the local marine boundary layer or in the regional troposphere. Atmospheric back trajectories may help in identifying time periods when there was continental air mass inflows. However, given that rain collections at Cape Point were done weekly, this presents a challenge in figuring out which time periods experienced an influx of polluted air masses. The overall information can be pieced together using hourly rain data and ancillary data from the site (for example, CO and Rn) which shed light on the origin of some of the air masses traveling to the site during the sampling period as they are, respectively, indicative of polluted and terrestrially-derived air masses. CO has proven to be an effective indicator of the larger scale redistribution of atmospheric pollutants by the long-range transport of polluted air masses (Duflot et al., 2010) and correlations between CO and Hg0 for polluted air masses have been documented in some locations (Jaffe et al., 2005) but not overall (e.g. Brunke et al., 2012). Many processes (oxidation of methane and other biogenic hydrocarbons, fossil fuel combustion, biofuel use and biomass burning) produce CO and its lifetime of weeks to months makes it an interesting tracer for studying atmospheric transport. Biomass burning is a leading pollution source in the tropics and more biomass burning occurs in Africa than any other continent (Hao and Liu, 1994). Therefore CO can also represent either a terrestrial or a pollution signal at Cape Point. In contrast, 99% of the 222 Rn in the atmosphere comes from rocks and minerals in the earth’s crust where it is dispersed by diffusion and advection after its production by decay of radium. Atmospheric 222Rn which has a half-life of 5.53 days (and decays to 210Pb) is not removed by either physical or chemical means and is dispersed sufficiently throughout the atmosphere to be a regional tracer (Baskaran, 2011). The mean lifetime of 222Rn is generally comparable to the time it takes air masses to move across continents and/or oceans but is short compared to the mixing time scale of the atmosphere. As a result 222Rn has been used across the globe as a signature for landderived air masses. The 222Rn detector at Cape Point was installed in 1999 and the efficacy of using those measurements in conjunction with wind direction and CO to classify air masses, either as maritime, continental or mixed has previously been demonstrated (Brunke et al., 2004, 2012). For Cape Point, 222Rn levels below 100 mBq m3 are considered to be typical of maritime air while

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

levels between 100 and 250 mBq m3 have some terrestrial influences, and higher values represent terrestrial air masses (Brunke et al., 2004). For CO, levels about 100 ppb or strong peaks in the signal are indicative of pollution inputs. Ancillary data was used to try and deduce the origin of air masses to examine the potential for terrestrial air masses to impact the sampling site during a rain event. Two examples are shown which indicate the difficulty in making such assignments given the nature of the Hg data e weekly total values, but indicate the potential for terrestrial inputs to this location. They also indicate that the air flow around the sampling site is complex and changes direction and source relatively rapidly. Firstly, consider the week of 26 Maye3 June 2008, where the rainfall concentration was 23.2 ng L1 and the flux 495 ng m2 wk1; i.e. a relatively high concentration and flux. Hourly rain data showed it rained on the 28, 29 and 30th of May, followed by a few dry days and more rainfall on the 2nd and 3rd of June 2008. It rained between 9.00 pm and 1.00 am from 2nd June 2008 into the early hours of 3 June 2008 and there was more rain at 11.00 am and again at 7.00 pm on 3 Jun 2008. During that period 222Rn values ranged from 54 to 5239 mBq m3, with a median of 705 mBq m3. Values <200 mBq m3 were only recorded from around 12 noon to midnight on 3 June 2008 (Fig. 6a). This suggests that air masses were of non-marine origin most of the time during that rainfall. Brunke et al. (2004) showed that 222Rn

295

and CO strongly correlate when air masses pass over terrestrial regions with strong CO sources. However, in some instances these species strongly diverge, for example instances of high 222Rn/low CO levels may occur when air masses pass over sparsely populated regions with few CO sources such as the granite/gneiss region in the north Western Cape. Conversely, low 222Rn/high CO concentrations are encountered when maritime air moves across the greater Cape Town area that has many anthropogenic sources such that by the time it reaches Cape Point, the terrestrial contribution is negligible (Brunke et al., 2004). Back trajectories support the CO/Rn data (Fig. 6b). The 72-h back trajectory ending at 1400 UTC (1600 local time) on 3 June 2008 showed that over the previous 3 days air masses had traveled from a height of 1500e2000 m from areas to the east of Cape Point. After the passage of the rain event, winds switched to more westerly flow as indicated by the 72 h back trajectories arriving at Cape Point on Jun 6. Wet weather in this region during winter normally arrives with storms from the south and this is suggested by the wind directions and change during the passage of the rain event on June 2 and 3. Prior to the storm event, the more easterly winds were likely more impacted by anthropogenic sources, as suggested by the CO and Rn data (Fig. 6a) and became more marine during the rain event. Our evaluation suggests that rain collected during the week of 26 Maye3 June 2008 was subject to anthropogenic influences,

Fig. 6. a) Carbon monoxide (CO, ppb) and radon levels (Rn, mBq m3) from midday June 1 to midnight June 6, 2008 for Cape Point (labels indicate midday); and b) 72 h back trajectories generated using the NOAA HYSPLIT Model for Cape Point ending at 1400 UTC on 3 June 2008 and on June 6, 2008. Back trajectories derived using NOAA software, and incorporated GDAS meteorological data, by user on Tuesday March 19, 2013, and for heights of 500, 1000 and 1500 m above ground level (AGL). Reprinted with permission from NOAA (Draxler and Rolph, 2013).

296

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

Fig. 7. a) Carbon monoxide (CO, ppb) and radon levels (Rn, mBq m3) from midday June 30 to midnight July 7, 2008 for Cape Point (labels indicate midday); and b) 72 h back trajectories generated using the NOAA HYSPLIT Model for Cape Point ending at 1400 UTC on 3 July and July 7, 2008. Back trajectories derived using NOAA software, and incorporated GDAS meteorological data, by user on Tuesday March 19, 2013, and for heights of 500, 1000 and 1500 m above ground level (AGL). Reprinted with permission from NOAA (Draxler and Rolph, 2013).

and this is both reflected in the relatively high concentration and flux for this week. It is worth noting that Brunke et al. (2012) used the trace gas signals at Cape Point from identified pollution events to reconstruct the emissions inventory for South Africa, further confirming the importance of pollution inputs as source for Hg and metals at Cape Point. They identified polluted air masses arriving from the east, which likely reflects local inputs from the industrial sector in the Western Cape region which is concentrated east of the city center, and from the informal housing, such as Khayelitsha, where heating in winter relies on coal and biomass burning. We also generated trajectories for the week of 30 June to 7 July 2008, where the THg concentration was 10.7 ng L1 but where the flux (450 ng m2 wk1) was comparable to the previous example. During this time period the rain occurred from 3 to 7 July 2008. A 72-h back trajectory ending on 3 July 2008 (Fig. 7b) shows some influence from the continent and the northwesterly flow characteristics of a storm and again the winds switch to more marine sources over time. Overall, 222Rn values were elevated from 1 to 3 July 2008 peaking at 6059 mBq m3 on the night of 1st July 2008; with a minimum of 1107 mBq m3 on 3 July 2008 and a median value of 2395 mBq m3 over the whole period (Fig. 7a), suggesting that the longer-term origin of the air was not totally marine. This indicates terrestrial influences especially as values were over 5000 mBq m3 (Brunke et al., 2004). Even given that there is a seasonal cycle with elevated 222Rn values (>1000 mBq m3) during the winter months of April to August at Cape Point, these values are elevated. During this time CO values ranged from 58 to 227 ppb

(background levels are w45 ppb; Brunke et al., 2010) showing elevated peaks in concentration over the period (Fig. 7a). Overall, these two examples of sources of air masses during precipitation, and other similar situations which are documented in the Rn and CO data (Brunke et al., 2010) demonstrate that Cape Point is subject to both marine and continental air masses and is not a background location, especially during the majority of the wet season. Sectored sampling of air removes such biases but this was not possible with the wet deposition samples collected here. However, while these results may not provide a clear signal of the regional background deposition at this site, they do indicate the ranges in concentrations and fluxes to the region, and the potential importance of anthropogenic inputs to deposition at Cape Point. Clearly, more data is needed to elucidate the chemical and atmospheric processes driving deposition at Cape Point, and associated events such as Hg depletion events. Acknowledgments This work was supported by the NSF International Program grant “Developing Global Scientists and Engineers through the Study of Mercury Environmental Issues in Southern Africa” and an NSF Chemical Oceanography grant “Air-Sea Exchange and Boundary Layer Chemistry of Mercury over the Open Ocean”, which partially supported Susan’s PhD research. We would like to thank Vernon Somerset, Andreas Trüe and the Council for Scientific and Industrial Research (CSIR), and Ernst Brunke, Casper Labuschagne and the South African Weather Service for help with sample

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

collection, storage and shipping. We also thank Ernst Brunke and colleagues for help with site setup and for provision of ancillary data for Cape Point (CO and 222Rn). The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://ready.arl.noaa.gov) used in this publication. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.atmosenv.2013.04.009. References Archived Weather Data. (accessed October, 2012.). www.tutiempo.net. Baskaran, M., 2011. Po-210 and Pb-210 as atmospheric tracers and global atmospheric Pb-210 fallout: a review. Journal of Environmental Radioactivity 102, 500e513. Batterman, S., Su, F.-C., Jia, C., Naidoo, R.N., Robins, T., Naik, I., 2011. Manganese and lead in children’s blood and airborne particulate matter in Durban, South Africa. Science of the Total Environment 409, 1058e1068. Brunke, E.-G., Labuschagne, C., Parker, B., Scheel, H.E., Whittlestone, S., 2004. Baseline air mass selection at Cape Point, South Africa: application of 222Rn and other filter criteria to CO2. Atmospheric Environment 38, 5693e5702. Brunke, E.-G., Labuschagne, C., Ebinghaus, R., Kock, H.H., Slemr, F., 2010. Total gaseous mercury depletion events observed at Cape Point during 2007e2008. Atmospheric Chemistry and Physics 10, 1121e1131. Brunke, E.-G., Ebinghaus, R., Labuschagne, C., Kock, H.H., Slemr, F., 2012. Emissions of mercury in southern Africa derived from long-term observations at Cape Point, South Africa. Atmospheric Chemistry and Physics 12, 7465e7474. Chester, R., 2003. Marine Geochemistry, second ed. Blackwell Science, Malden. Cho, S.-H., Richmond-Bryant, J., Thornburg, J., Portzer, J., Vanderpool, R., Cavender, K., Rice, J., 2011. A literature review of concentrations and size distributions of ambient airborne Pb-containing particulate matter. Atmospheric Environment 45, 5005e5050. Conaway, C.H., Black, F.J., Weiss-Penzias, P., Gault-Ringold, M., Flegal, A.R., 2010. Mercury speciation in Pacific coastal rainwater, Monterey Bay, California. Atmospheric Environment 44, 1788e1797. Corbitt, E.S., Jacob, D.J., Holmes, C.D., Streets, D.G., Sunderland, E.M., 2011. Global source-receptor relationships for mercury deposition under present-day and 2050 emissions scenarios. Environmental Science and Technology 45, 10477e 10484. Dabrowski, J.M., Ashton, P.J., Murray, K., Leaner, J.J., Mason, R.P., 2008. Anthropogenic mercury emissions in South Africa: coal combustion in power plants. Atmospheric Environment 42, 6620e6626. Draxler, R.R., Rolph, G.D., 2013. HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory). NOAA Air Resources Laboratory, Silver Spring, MD. Model access via NOAA ARL READY Website. http://ready.arl.noaa.gov/HYSPLIT.php. Duflot, V., Dils, B., Baray, J.L., De Mazière, M., Attié, J.L., Vanhaelewyn, G., Senten, C., Vigoroux, C., Clain, G., Delmas, R., 2010. Analysis of the origin of the distribution of CO in the subtropical southern Indian Ocean in 2007. Journal of Geophysical Research 115, D22106. Dutt, U., Nelson, P.F., Morrison, A.M., Strezov, V., 2009. Mercury wet deposition and coal-fired power station contributions: an Australian study. Fuel Processing Technology 90, 1354e1359. Ebinghaus, R., Jennings, S.G., Schroeder, W.H., Berg, T., Donaghy, T., Guentzel, J., Kenny, C., Kock, H.H., Kvietkus, K., Landing, W., Muhleck, T., Munthe, J., Prestbo, E.M., Schneeberger, D., Slemr, F., Sommar, J., Urba, A., Wallschlager, D., Xiao, Z., 1999. International field inter-comparison measurements of atmospheric species at Mace Head, Ireland. Atmospheric Environment 33, 3063e 3073. Engle, M.A., Tate, M.T., Krabbenhoft, D.P., Kolker, A., Olson, M.L., Edgerton, E.S., DeWild, J.F., McPherson, A.K., 2008. Characterization and cycling of atmospheric mercury along the central US Gulf Coast. Applied Geochemistry 23, 419e437. Guentzel, J.L., Landing, W.M., Gill, G.A., Pollman, C.D., 1995. Atmospheric deposition of mercury in Florida: the FAMS project (1992e1994). Water Air and Soil Pollution 80, 393e402. Guentzel, J.L., Landing, W.M., Gill, G.A., Pollman, C.D., 2001. Processes influencing rainfall deposition of mercury in Florida. Environmental Science and Technology 35, 863e873. Guo, Y., Feng, X., Li, Z., He, T., Yan, H., Meng, B., Zhang, J., Qiu, G., 2008. Distribution and wet deposition fluxes of total and methyl mercury in Wujiang River Basin, Guizhou, China. Atmospheric Environment 42, 7096e7103. Hall, B.D., Manolopoulos, H., Hurley, J.P., Schauer, J.J., St. Louis, V.L., Kenski, D., Graydon, J., Babiarz, C.L., Cleckner, L.B., Keeler, G.J., 2005. Methyl and total mercury in precipitation in the Great Lakes region. Atmospheric Environment 39, 7557e7569. Halstead, M.J.R., Cunnighame, R.G., Hunter, K.A., 2000. Wet deposition of trace metals to a remote site in Fiordland, New Zealand. Atmospheric Environment 34, 665e676.

297

Hao, W.M., Liu, M.-H., 1994. Spatial and temporal distribution of tropical biomass burning. Global Biogeochemical Cycles 8, 495e503. Herrera, J., Rodriguez, S., Baez, A.P., 2009. Chemical composition of bulk precipitation in the metropolitan area of Costa Rica, Central America. Atmospheric Research 94, 151e160. Holmes, C.D., Jacob, D.J., Mason, R.P., Jaffe, D.A., 2009. Sources and deposition of reactive gaseous mercury in the marine atmosphere. Atmospheric Environment 43, 2278e2285. Iverfeldt, A., Munthe, J., 1993. Determining the wet deposition of mercuryda comparison of weekly, biweekly, and monthly collection of precipitation samples. Measurement of toxic and related air pollutants. In: Proceedings of the 1993 US EPA/A&WMA International Symposium. Jaffe, D., Prestbo, E., Swartzendruber, P., Weiss-Penzias, P., Kato, S., Takami, A., Hatakeyama, S., Kajii, Y., 2005. Export of atmospheric mercury from Asia. Atmospheric Environment 39, 3029e3038. Kading, T.J., Mason, R.P., Leaner, J.J., 2009. Mercury contamination history of an estuarine floodplain reconstructed from a 210Pb-dated sediment core (Berg River, South Africa). Marine Pollution Bulletin 59, 116e122. Keeler, G.J., Gratz, L.E., Al-Wali, K., 2005. Long-term atmospheric mercury wet deposition at Underhill, Vermont. Ecotoxicology 14, 71e83. Lacerda, L.D., Paraquetti, H.H.M., Rezende, C.E., Silva, L.F.F., Silva, E.V., Marins, R.V., Ribeiro, M.G., 2002. Mercury concentrations in bulk atmospheric deposition over the coast of Rio de Janeiro, Southeast, Brazil. Journal of the Brazilian Chemical Society 13, 165e169. Lamborg, C.H., Rolfhus, K.R., Fitzgerald, W.F., 1999. The atmospheric cycling and airsea exchange of mercury species in the south and equatorial Atlantic Ocean. Deep-Sea Research II 46, 957e977. Lamborg, C., Hammerschmidt, C.R., Gill, G.A., Mason, R.P., Gichuki, S., 2012. An intercomparison of procedures for the determination of total mercury in seawater and recommendations regarding mercury speciation during GEOTRACERS cruises. Limnology and Oceanography Methods 10, 90e100. Landing, W.M., Perry, J.J., Guentzel, J.L., Gill, G.A., Pollman, C.D., 1995. Relationship between the atmospheric deposition of trace elements, major ions and mercury in Florida: the FAMS project (1992e1993). Water Air and Soil Pollution 80, 343e352. Landis, M.S., Stevens, R.E., Schaedlich, F., Prestbo, E.M., 2002. Development and characterization of an annular denuder methodology for the measurement of divalent inorganic reactive gaseous mercury in ambient air. Environmental Science and Technology 36, 3000e3009. Laurier, F., Mason, R.P., 2007. Mercury concentration and speciation in the coastal and open ocean boundary layer. Journal of Geophysical Research-Atmospheres. Art# D06302. Leaner, J.J., Dabrowski, J.M., Mason, R.P., Resane, T., Richardson, M., Ginster, M., Gericke, G., Petersen, C.R., Masekoameng, E., Ashton, P.J., Murray, K., 2009. Mercury emissions from point sources in South Africa. In: Pirrone, N., Mason, R.P. (Eds.), Mercury Fate and Transport in the Global Atmosphere: Emissions, Measurements and Models. Springer, pp. 113e130 Chapter 5. Lombard, M.A.S., Bryce, J.G., Mao, H., Talbot, R., 2011. Mercury deposition in southern New Hampshire, 2006e2009. Atmospheric Chemistry and Physics 11, 7657e7668. Masekoameng, K.E., Leaner, J.J., Dabrowski, J., 2010. Trends in anthropogenic mercury emissions estimated for South Africa during 2000-2006. Atmospheric Environment 44, 3007e3014. Mason, R.P., Sheu, G.-R., 2002. The role of the ocean in the global mercury cycle. Global Biogeochemical Cycles 16. Article #1093. Mason, R.P., Lawson, N.M., Sheu G-, R., 2000. Annual and seasonal trends in mercury deposition in Maryland. Atmospheric Environment 34, 1691e1701. Mason, R.P., Lawson, N.M., Sheu G-, R., 2002. The urban atmosphere: an important source of trace metals to nearby waters? In: Lipnick, R.L., Mason, R.P., Phillips, M.L., Pittman Jr., C.U. (Eds.), Chemicals in the Environment: Fate, Impacts, and Remediation, ACS Symposium Series No. 806. American Chemical Society Publ., pp. 203e222. Naicker, K., Cukrowska, E., McCarthy, T.S., 2003. Acid mine drainage arising from gold mining activity in Johannesburg, South Africa and environs. Environmental Pollution 1221, 29e40. Pacyna, E.G., Pacyna, J.M., Steenhuisen, F., Wilson, S., 2006. Global anthropogenic mercury emission inventory for 2000. Atmospheric Environment 40, 4048e4063. Papu-Zamxaka, V., Mathee, A., Harpham, T., Barnes, B., Roellin, H., Lyons, M., Jordaan, W., Cloete, M., 2009. Elevated mercury exposure in communities living alongside the Inanda Dam, South Africa. Journal of Environmental Monitoring 12, 472e477. Pirrone, N., Cinnirella, S., Feng, X., Finkelman, R.B., Friedli, H.R., Leaner, J.J., Mason, R.P., Muykherjee, A.B., Stracher, J.B., Streets, D.G., Telmer, K., 2010. Global mercury emissions to the atmosphere from anthropogenic and natural sources. Atmospheric Chemistry and Physics 10, 5951e5964. Sakata, M., Asakura, K., 2009. Factors contributing to seasonal variations in wet deposition fluxes of trace elements at sites along Japan Sea coast. Atmospheric Environment 43, 3867e3875. Sakata, M., Marumoto, K., Narukawa, M., Asakura, K., 2006. Regional variations in wet and dry deposition fluxes of trace elements in Japan. Atmospheric Environment 40, 521e531. Schmeltz, D., Evers, D.C., Driscoll, C.T., Artz, R., Cohen, M., Gay, D., Haeuber, R., Krabbenhoft, D.P., Mason, R.P., Morris, K., Weiner, J.G., 2011. MercNet: a national mercury monitoring network to assess responses to changing mercury emissions in the United States. Ecotoxicology 20, 1713e1725.

298

S.W. Gichuki, R.P. Mason / Atmospheric Environment 79 (2013) 286e298

Sekonya, K.G., 2009. Characterisation of Ambient Atmospheric Aerosols Using Accelerator-Techniques. MS thesis. University of the Witwatersrand, Johannesburg, p. 106. Seo, Y.-S., Han, Y.-J., Choi, H.-D., Holsen, T.M., Yi, S.-M., 2012. Characteristics of total mercury (TM) wet deposition: scavenging of atmospheric mercury species. Atmospheric Environment 49, 69e76. Slemr, F., Brunke, E.-G., Labuschagne, C., Ebinghaus, R., 2008. Total gaseous mercury at the Cape Point G.A.W station and their seasonality. Geophysical Research Letters 35, L11807. Slemr, F., Brunke, E.-G., Ebinghaus, R., Kuss, F., 2011. Worldwide trend of atmospheric mercury since 1995. Atmospheric Chemistry and Physics 11, 2355e 2375. Soerensen, A.L., Sunderland, E.M., Holmes, C.D., Jacob, D.J., Yantosca, R.M., Skov, H., Christensen, J.H., Strode, S.A., Mason, R.P., 2010. An improved global model for air-sea exchange of mercury: high concentrations over the North Atlantic. Environmental Science and Technology 44, 8574e8580. Song, F., Gao, Y., 2009. Chemical characteristics of precipitation at metropolitan Newark in the U.S. East Coast. Atmospheric Environment 43, 4903e4913. Sprovieri, F., Hedgecock, I.M., Pirrone, N., 2010. An investigation of the origins of reactive gaseous mercury in the mediterranean marine boundary layer. Atmospheric Chemistry and Physics 10, 3007e3014. Streets, D.G., Devane, M.K., Lu, Z., Bond, T.C., Sunderland, E.M., Jacob, D.J., 2011. Alltime releases of mercury to the atmosphere from human activities. Environmental Science and Technology 45, 10485e10491.

Thorpe, A., Harrison, R.M., 2008. Sources and properties of non-exhaust particulate matter from road traffic: a review. Science of the Total Environment 400, 270e282. Vanarsdale, A., Weiss, J., Keeler, G., Miller, E., Boulet, G., Brulotte, R., Poissant, L., 2005. Patterns of mercury deposition and concentration in northeastern North America (1996e2002). Ecotoxicology 14, 37e52. Vuai, S.A.H., Tokuyama, A., 2011. Trend of trace metals in precipitation around Okinawa Island, Japan. Atmospheric Research 99, 80e84. Walters, C.R., Somerset, V.S., Leaner, J.J., Nel, J.C., 2011. A review of mercury pollution in South Africa: current status. Journal of Environmental Science and Health, Part A: Toxic/Hazardous Substances and Environmental Engineering 46, 1129e 1137. Wang, Y.M., Wang, D.Y., Meng, B., Zhao, L., Zhu, J.S., 2012. Spatial and temporal distributions of total and methylmercury in precipitation in core urban areas, Chongqing, China. Atmospheric Chemistry and Physics 12, 9417e9426. Willey, J.D., Inscore, M.T., Kieber, R.J., Skrabal, S.A., 2009. Manganese in coastal rainwater:speciation, photochemistry and deposition to seawater. Journal of Atmospheric Chemistry 62, 31e43. Zhang, L.M., Wright, L.P., Blanchard, P., 2009. A review of current knowledge concerning dry deposition of atmospheric mercury. Atmospheric Environment 43, 5853e5864. Zhang, X., Siddiqi, Z., Song, X., Mandiwana, K., Yousaf, M., Lu, J., 2012. Atmospheric dry and wet deposition of mercury in Toronto. Atmospheric Environment 50, 60e65.