Long-term trends in atmospheric concentrations of sulfate, total sulfur, and trace elements in the northeastern United States

Long-term trends in atmospheric concentrations of sulfate, total

sulfur, and trace elements in the northeastern United States

Liaquat Husain,1 Pravin P. Parekh, Vincent A. Dutkiewicz,1 Adil R. Khan, Karl Yang,

and Kamal SwamiWadsworth Center, New York State Department of Health, Albany, New York, USA

Received 7 April 2004; revised 14 June 2004; accepted 12 July 2004; published 18 September 2004.

[1] Concentrations of K, Sc, Mn, Fe, Zn, As, Se, Sb, Hg, and Pb were determined inquarterly composites of daily aerosol samples collected at Mayville, and 530 km downwindat Whiteface Mountain (1.5 km altitude), New York, for �20 years. SO4 concentrations[SO4] were determined in individual daily samples. Continuous hourly SO2 data arealso available for much of the period. [SO4] at Mayville were twice that at WhitefaceMountain, and total S (S as SO2 + SO4) burden was fourfold higher at Mayville. From 1979through 2002, [SO4] decreased by 59% atWhitefaceMountain, and atMayville the decreasewas 30% from 1984 to 2002. From 1979 to 2002, SO2 emissions in eight states upwind ofand contiguous with New York State (Ohio, Pennsylvania, Indiana, Illinois, Wisconsin,Kentucky,West Virginia, and Ontario, Canada) decreased by 49%. A linear relationship wasobserved between atmospheric [SO4] and [total S] burden at the two sites with thecumulative SO2 emissions. These observations suggest that any further reductions in SO2

emissions would result in a proportional decrease in [SO4] and [total S] across New YorkState and possibly across the northeastern United States. The data at WhitefaceMountain suggest that beginning in 1997, the decrease in [SO4] and [total S], relative to SO2

emissions, may be faster than the earlier period. Like [SO4] and [total S], the trace elementconcentrations were twofold to fivefold higher at Mayville than atWhitefaceMountain. Theconcentrations at both sites showed an unmistakable decrease over time. The largestdecreases were observed for Hg (16%/year at Whiteface Mountain and 10%/year atMayville) and Pb (14%/year at Whiteface Mountain and 10%/year at Mayville). Theremaining elements (except Sb), including the crustal elements K,Mn, Sc, and Fe, showed adecreases of 3–5%/year. Trends for Sb atWhitefaceMountain and forMn atMayville couldnot be accurately discerned, apparently due to some nearby emissions. Apparently, thereductions in the emissions of SO2 and particulate matter have also resulted in the decreaseof atmospheric burden of trace elements and an improvement in air quality. INDEX TERMS:

0305 Atmospheric Composition and Structure: Aerosols and particles (0345, 4801); 0345 Atmospheric

Composition and Structure: Pollution—urban and regional (0305); 0365 Atmospheric Composition and

Structure: Troposphere—composition and chemistry; 0368 Atmospheric Composition and Structure:

Troposphere—constituent transport and chemistry; KEYWORDS: aerosol sulfate, acid rain, long-term trends,

tropospheric composition, atmospheric transport, trace elements

Citation: Husain, L., P. P. Parekh, V. A. Dutkiewicz, A. R. Khan, K. Yang, and K. Swami (2004), Long-term trends in atmospheric

concentrations of sulfate, total sulfur, and trace elements in the northeastern United States, J. Geophys. Res., 109, D18305,

doi:10.1029/2004JD004877.

1. Introduction

[2] Recently, we have shown that concentrations ofatmospheric SO4 and total sulfur (sum of sulfur present asSO2 and SO4) have decreased across New York State from1979 to 1998. We also showed that the measured SO4

concentrations were linearly related to the SO2 emissions

in the eight states upwind of and contiguous with NewYork state [Husain et al., 1998; Dutkiewicz et al., 2000].The primary source of SO2 in those states is coal com-bustion. In addition to SO2, coal burning emits coal flyash, bearing trace elements as well as volatile elements asvapors. The latter, upon condensation, form particles.Aerosols bearing trace elements are also introduced intothe atmosphere through a myriad of industrial operations,transportation, and other human activities [e.g., Gone etal., 2000, and references therein]. Entrainment of soilparticles, forest fires, and sea salt are examples of naturalsources in the region. The prevailing westerly winds

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, D18305, doi:10.1029/2004JD004877, 2004

1Also at Department of Environmental Health and Toxicology, Schoolof Public Health, State University of New York, Albany, New York, USA.

Copyright 2004 by the American Geophysical Union.0148-0227/04/2004JD004877$09.00

D18305 1 of 12

transport SO2, aerosol SO4, and the trace elements overdistances of hundreds of km in the northeastern UnitedStates [Husain and Samson, 1979; Husain et al., 1984;Dutkiewicz et al., 1987; Rahn and Lowenthal, 1984,1985]. The aerosols degrade visibility and air quality.The particulate matter that is less than 2.5 mm in diameter,PM2.5, is considered to have serious impacts on humanhealth. The regulatory controls implemented over the lastquarter century to decrease SO2 and particulate emissionsshould result in decreases in particulate matter and theassociated trace elements. No long-term data on traceelement concentrations in the atmosphere of the Northeastare available. We have therefore undertaken this study toinvestigate the temporal changes in the concentrations ofselected trace elements across the northeastern UnitedStates. Aerosol samples collected daily at WhitefaceMountain from 1979 to 2001 and from 1984 to 2001 atMayville, New York were analyzed. The sites are 530 kmapart, in rural areas devoid of industry, and hence wellsuited for the study of the transport of trace elements, SO2,and SO4. The wind flow, especially during summer, issuch that the air masses traveling through the midwesternUnited States usually pass through Mayville and on toWhiteface Mountain. Therefore the data from these sitesshould enable a study of (1) long-term changes in SO4 andtrace element concentrations, and (2) the effect of SO2

reductions, and other ameliorations of industrial operationsand implementation of pollution controls during the last2 decades, on SO4 and trace element concentrations.

2. Experimental Procedure

2.1. Sample Collection and Pellet Making

[3] Details of sample collection have been publishedearlier [Husain et al., 1998]. Briefly, the aerosols werecollected daily using high-volume samplers on 25 cm �20 cm Whatman 41 filters. SO4 concentrations were deter-mined in the daily samples using ion chromatography (IC)by the procedure described earlier [Husain et al., 1984]usually within a few months. The total number of samplescollected was in excess of 10,000. To analyze each sampleindividually for trace elements was beyond our resourcesand not necessary for the goals of this study. We thereforedecided to prepare composite samples from our archivedfilters for each quarter of a given sampling year, whichshould suffice for the study of both seasonal and annualtrends. From each daily filter, a 0.64 cm disc was punchedwith a high-purity punch designed to minimize trace elementcontamination. Samples were divided into quarters as1 January through 31 March, 1 April through 30 June, 1 Julythrough 30 September, and 1 October through 31 December.For trace element determination by instrumental neutronactivation analysis (INAA) the composite samples from eachquarter were pelletized using a hydraulic press. For certainyears, instead of the usual 24 hour sampling, short-termsampling (2 to 6 h) was carried out, especially duringsummers. Pellets obtained for these years were relativelythicker due to the number of samples which exceeded 90 perquarter. Pellets of three different thicknesses had to beprepared. Three multielement standards were prepared bydepositing solutions of known concentrations on pellets ofWhatman 41 filters of the same thickness as the samples.

Two National Institute of Standard and Testing’s StandardReference Materials (NIST-SRM) were used as qualitycontrol: Oyster Tissue NIST-SRM 1566 and Orchard LeavesNIST-SRM 1571. In addition, duplicate pellets were pre-pared for all four quarters from both sites to help establishuncertainties.

2.2. Instrumental Neutron Activation Analysis

[4] The quarterly composite samples were analyzed forK, Sc, Fe, Zn, As, Se, Sb and Hg using INAA. The detailsof INAA have been given elsewhere [Dutkiewicz et al.,1987]. The g spectra were collected using Ge and GeLidetectors connected to a Genie 2000 control system(Canberra Industries). Because of the thickness of thesesamples, they were flipped at the half-way point ofcounting to minimize any bias due to sample heterogene-ity. The spectra were stored on disc, and g-ray lineintensities were determined offline with the Genie 2000 g

analysis software. Final concentrations were determinedthrough comparison to appropriate known standards, afterbackground and decay correction, with a computer codethat was developed in-house. Only samples collectedthrough 1999 were analyzed by INAA.

2.3. Inductively Coupled Plasma Mass Spectrometry(ICP-MS) Analysis

[5] As INAA is nondestructive, after the pellets haddecayed to near background Mn, Fe, Zn, As, Se, Sb, andPb were determined by ICP-MS, after dissolution of thesamples by microwave oven digestion [Swami et al., 2001;Yang et al., 2002], using a Hewlett Packard 4500 Induc-tively Coupled Plasma Mass Spectrometer, equipped withChemStation Software and a Cetac Technologies ASX 500autosampler. Concentrations of Mn and Pb were deter-mined only by ICP-MS, K, Sc, Sb, and Hg only by INAA,and Fe, Zn, As and Se by both techniques. The ICP-MSand INAA data for these species were in excellent agree-ment and hence the average of the two methods were used.After 1999 only ICP-MS was used. While K, Sc and Hgwere not available, Sb results by ICP-MS were used.NIST-SRM 1648 Urban Particulate Matter (UPM) wasused as a quality control. In the UPM, Mn, Fe, Zn, As,Sb and Pb showed better than 90% recovery, with relativestandard deviation less than 10%, but Se showed only70% recovery with relative standard deviation of 6%.However, it is likely that the recoveries of some elementson UPM may not represent the samples, which mayintroduce a substantially different matrix to the ICP-MS.In fact, our previous study showed nearly 90% agreementbetween INAA and ICP-MS results for Se on samplescollected at Whiteface Mountain and Mayville [Yang et al.,2002]. The duplicate composite samples were run by bothICP-MS and INAA. For cases where concentrations weregreater than the minimum reporting level, relative differ-ences ranged from 2 to 3% for As and Se by INAA to15% for Zn and Pb by ICP-MS, however, the relativedifferences for most elements fell between 7 to 9%. Wecombined the duplicate precision with a 7% samplinguncertainty to obtain an overall analytical uncertainties;10%, 13%, 10%, 11%, 14%, 8.3%, 9.6%, 8.6%, 19% and17%, respectively for K, Sc, Mn, Fe, Zn, As, Se, Sb, Hg,and Pb.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

2 of 12

D18305

[6] Sulfate was determined by ion chromatography on awater extract from a portion of each filter. The methodswere described in Husain et al. [1984]. Analytical uncer-tainties are ±7%.

3. Results and Discussion

3.1. Trends in Sulfate and Total Sulfur Concentrations

[7] In order to study the long-term trends in the atmo-spheric concentrations of chemical species, it is essential toconduct sampling and analysis over a few decades, aschanges are often small, a few percent or less per year. Thisstudy presents continuous data from 1 January 1979 to31 December 2002 at Whiteface Mountain and from1 January 1984 to 31 December 2002 at Mayville. In viewof the expectations that yearly changes would be very small,we have deliberately avoided making any changes in thesampling and analytical methodology over the duration ofthis study so as to avoid introducing systematic uncertainties

in the data that could mask long-term trends in concentra-tions. Therefore the same collection substrate (Whatman41 filter), samplers (high-volume), analytical methodology(IC, INAA, and ICP-MS) have been used for all samples. Theuse of 20 cm� 25 cm filters has provided ample samples for avariety of analyses. To the extent possible, we have alsosampled every day using either 24 h or 48 h samples.Although skip day sampling, i.e., every sixth or every thirdday, produces a reduced sample burden to analyze, the datamay not be as suitable to study the long-term trends.[8] While the uncertainty of a single sulfate measure-

ments is estimated at 7%, when data are averaged for trendanalysis the statistical uncertainty is decreased. This will beparticularly true in light of the efforts mentioned above tominimize sampling and analytical bias. To evaluate thisempirically we computed 3 year relative standard deviationsfor the SO4 and total sulfur measurements at both sites for1983 through 2000 as well as for the regional SO2 emis-sions that is defined later in the text. Limiting our analysisto the cases when the standard deviation of the regionalemissions was <±3% and accounting for the fact thattransport/meteorology can at times have a large confound-ing effect on measured concentrations, uncertainties in theannual mean for SO4 and total sulfur are conservativelyestimated at ±3.5%. The uncertainty for 3 year means istherefore ±2%. Applying a standard t-test and statisticaltables, differences on an annual basis of 7% or more aresignificant at the 95% confidence level. For 3 year means a5% difference is significant at the 95% confidence level.3.1.1. Sulfate[9] Monthly SO4 concentrations at Whiteface Mountain

from January 1979 to December 2002, as determined fromthe daily measured concentrations, are shown in Figure 1. Arecurring feature of the data is the annual summer peak.Concentrations are highest during the warm months (April–September) and lowest during the cold (October–March),peaking during July or August. The warm season peaksshow an unmistakable decrease in concentrations over time.The trend can be easily detected from the 3 year averages(Figure 1). Owing to missing data during 1980 and 1982 atWhiteface Mountain, annual means for 1979 and 1981 areshown separately, and we focus on the 3 year means for theperiod since 1983. At Whiteface Mountain, the concentra-tions decreased sharply from 1979 to 1981 but then onlyslightly through 1991. The mean concentration for the1992–1994 period was 20% lower, compared to the1981–1991 period. An additional 30% decrease was ob-served for 1995 to 1997. Since 1997, the concentrations haveremained constant. There was an unmistakable decrease ofmore than twofold over 2 decades. Concen trations at May-ville (Figure 1) showed a similar pattern to WhitefaceMountain with relatively steady concentrations between1984 and 1991, followed by a 13% decrease in 1992–1994, and an additional 22% decrease for 1995–2002. Thenet concentrations decreased by 31% for 1992–2002 period,relative to that during the 1980s.[10] Husain et al. [1998] demonstrated that the changes in

the annual mean SO4 concentrations at Whiteface Mountainand Mayville linearly followed changes in the SO2 emis-sions in eight states (PA, OH, IN, IL, WI, MI, KY, WV)upwind and contiguous with New York State. Data fromother networks in the Northeast have reported similar results

Figure 1. Monthly mean SO4 concentrations at Mayvilleand Whiteface Mountain, New York. Solid horizontal linesrepresent 3 year means (see text). Also shown are the annualmean SO2 emissions from the Midwest + Ontario (see text).The dashed curve is a smoothing spline added as a visualaid, and SO2 emissions are expressed in kilometric tons perday.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

3 of 12

D18305

[Malm et al., 2002; Hicks et al., 2002]. The SO2 emissionsfor the eight states plus Ontario, Canada, are shown inFigure 1. For simplification we refer to these as the regionalemissions. The states included here contribute around 40%of the US emissions total. As total statewide emissions datawere only available on the worldwide web through 2000,we used the 2001 and 2002 electric utility emissions andtotal emissions for 2000 to scale emissions through 2002.The electric utility emissions represented 69% of the na-tional total in 2001, so this approximation should be fairlyaccurate, considering the uncertainties that go into comput-ing the total emissions. The SO2 emissions so calculated areshown in Figure 1. The emissions were very steady from1982–1991, followed by a 22% decrease in 1995. Subse-quently, the emissions remained relatively constant through1999, with a 17.5% decrease through 2002. With the meanof 1982–1991 used as the baseline, we calculated thatemissions decreased by 39% in 2001–2002. Correspond-ingly, SO4 concentration have decreased by 30% at May-ville and 46% at Whiteface Mountain.[11] One way to view the relationship between SO4 and

SO2 emissions is to consider the trend in their ratio asshown in Figure 2. The lines are locally weighted leastsquares, and curves have been drawn to smooth the trendsthrough the data. At Mayville the curve suggests a constantratio at 0.200 ± 0.018 mg m�3 per kilometric ton of SO2

(KMT) day�1, until the year 2002. A one standard deviationin the mean ratio of 8.9% over 9 years suggests that thesource-receptor relationship has remained remarkably con-stant; and thus it may be used to predict future changes inSO4 concentrations in response to reduction in SO2 emis-sions. There is, however, a slight up turn in the ratio for2001 and 2002, to 0.239 and 0.230, respectively. This trendwill need to be monitored in the future. Although the ratiosat Whiteface Mountain are not as constant, there is muchless scatter than at Mayville, and a very marked trend isevident. On the basis of the weighted least squares curve,there are two distinct periods, with the transition occurringaround 1992. Prior to 1992 the ratio was 0.100 ± 0.009 andsince 1995 it has decreased to 0.082 ± 0.0087 mg m�3 per

KMT of SO2 day�1, an 18% change. During 1992–1994the ratios vary between these levels, while within eachperiod the one standard deviation of the ratios is <±10%.As at Mayville, the ratios for 2001 and 2002, 0.090 and0.093, respectively, are slightly above average. This trendshould also be watched. Since we have maintained the samefilters, sampling methods and analytical protocol, a system-atic measurement bias is not likely to be the explanation.Alternately, some change may be occurring in the spatialdistribution of the SO2 emissions. We have assumed that allsources in the selected region are equally effective atimpacting SO4 levels at our sites; distance and directionof the sources have not been factored in individually.Despite these approximations, the SO4 to SO2 emissionratios in Figure 2 show a very simple and systematicpattern. The data have established that (1) the atmosphericSO4 concentrations at Mayville have been twice as high asthose at Whiteface Mountain, and (2) SO4 concentrationsrelative to SO2 emissions may now be decreasing atWhiteface Mountain approximately 20% faster than during1981–1991.3.1.2. Total Sulfur[12] In addition to using SO4 concentrations measured at

Mayville and Whiteface Mountain, Husain et al. [1998] alsoevaluated the impact of changes in the regional SO2

emissions on the total S (sum of SO2 and SO4) at thesesites. SO2 in the atmosphere is continuously oxidized toSO4, and at the same time it is also dispersed as well asremoved so essentially all of the sulfur emitted from fossilfuel combustion is present as SO2 or SO4. Barring signif-icant contributions from local SO2 emissions, total S repre-sents a better accounting of regional emissions as onemoves away from the source. While differences in theSO2 and SO4 removal rates may impact the spatial relation-ship between total S and SO2 emissions, it should not have asignificant impact on the temporal trend at a given site,however, changes in SO2 emissions from nearby sourcesmay be important. Year-round measurements of SO2 beganat Whiteface Mountain in 1988 and at Westfield, 16 kmnorthwest of Mayville, in 1991. SO2 was continuouslymonitored with TECO model 43 and 43S pulsed fluores-cence analyzer by New York State Department of Environ-mental Conservation (NYS DEC, private communication).While the Mayville and Westfield sites are relatively rural,there is a coal-burning power plant at Dunkirk, around30 km to the northeast of both. Trajectory calculationsindicate that air flow from Dunkirk toward Westfield israre. Thus on an annual basis, its impact should be minimal.[13] Monthly and 3 year average total S concentrations

for both Whiteface Mountain and Mayville are given inFigure 3. On an annual basis SO4 contributes from 15 to17% of the total S at Mayville and from 21 to 34% atWhiteface Mountain. Thus total S has a significant compo-nent that is independent of SO4 measurements. Like SO4,total S in Figure 3 varies with time of the year, however, thehighest concentrations occur during the cold rather than thewarm months. In addition, total S concentrations at bothsites were higher prior to 1995. For 1992–1994 total Saveraged 7.84 ppb at Mayville and 1.87 ppb at WhitefaceMountain while the 1995–1997 means were 5.92, and1.34 ppb, respectively. That is a 24% decrease at Mayvilleand a 28% decrease at Whiteface Mountain. These changes

Figure 2. Annual ratios of SO4 at Mayville and WhitefaceMountain to upwind SO2 emissions (see text). Solidhorizontal lines show mean values (see text), and curvesare smoothing splines added as a visual aid. Dashedhorizontal lines represent a range of ±10% about the means.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

4 of 12

D18305

are similar to that for regional SO2 emissions in Figure 1.The means for 1998–2000 at both sites are not significantlydifferent to the 1995–1997, however, for 2001–2002 therewas a 12% decrease at Mayville compared to a 24%increase at Whiteface Mountain. Both are significant atthe 95% confidence level. The regional SO2 emissions inFigure 1 are consistent with the Mayville trend suggestingthat either a source or sources between Mayville andWhiteface Mountain have significantly increased SO2 emis-sions in 2001–2002. At this time the source is unknown.[14] The ratios of total S to the regional SO2 emissions

for both sites are plotted in Figure 4. At Mayville,the mean ratio was 0.32 ± 0.02 compared to 0.079 ±0.013 ppb/KMT day�1 at Whiteface Mountain. Thus theatmospheric concentration of total S at Mayville, 530 kmupwind of Whiteface Mountain, is fourfold higher. The onestandard deviation of the ratio is only ±6% at Mayville and±16% at Whiteface Mountain. A large part of this deviationoccurs in 2001 and 2002, when the ratios are 22% abovethe average value. If only 1988 through 2000 data areconsidered, the ratio is 0.076 ± 0.001 ppb/KMT day�1, ora deviation of ±12.6%. With the exception of 2001and 2002, there was no significant change in totalS concentrations relative to SO2 emissions, i.e., their ratioshave remained constant across the state over 2 decades.Hence it is reasonable to conclude that, at least in the nearfuture, reductions in upwind regional SO2 emissions willresult in a proportional decrease in atmospheric totalS burden in the northeastern United States.

3.2. Trends in Trace Elements Concentrations

[15] Annual mean concentrations were determined byaveraging the quarterly composite measurements. The read-er may acquire the quarterly data by directly writing to theauthors. The annual mean concentrations along with uncer-

tainties are given in Table 1 for Whiteface Mountain from1979 to 2001, and at Mayville from1984 to 2001. The Hgmeasurements from INAA are included inspite of the higheruncertainties, and the fact that more than 95% of atmo-spheric Hg occurs in gas phase (Hgo). Because of the shortatmospheric lifetime of particulate Hg as compared togaseous, wet and dry deposition of particulate Hg may bethe primary pathway of atmospheric Hg into water systemswhere it can bioaccumulate in fish [Olmez et al., 1998].Additionally, there is a serious paucity of long-term Hg dataand the impact that pollution reduction measures on coalburning have on Hg is not yet well understood.3.2.1. Grouping Elements by Source[16] A simple way to determine whether a chemical

element in the atmosphere is derived from soil or from

Figure 3. Monthly mean total S concentrations at Whiteface Mountain and Mayville, New York. Solidcurves are the result of three-point running means, while the horizontal lines are 3 year means.

Figure 4. Same as Figure 2, except for total S.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

5 of 12

D18305

anthropogenic emissions is to determine its enrichment inaerosols compared to soil. The Enrichment factor (EF) foran element A can be defined as

EF ¼ A=Xð Þsample= A=Xð Þcrust;

where (A) represents the concentration of A, and X is theconcentration of a crustal reference element, usually Al, Fe,or Sc. Aluminum concentrations were not measured in thiswork, and we prefer to use Sc over Fe because the steelindustry may contribute a significant amount of Fe. Thenumerator is the ratio of A/X in the sample while thedenominator is the ratio in the crust. We used crustalabundances fromMason [1966] to determine the EFs for theelements analyzed in this work. The mean values and onestandard deviations for the entire sampling period are givenin Table 2 for Mayville and Whiteface Mountain. The EFsfor K, Mn, and Fe are very similar at the two sites. Ideally, a

purely crustal element has an EF of 1, so an EF of 2 wouldindicate equal contributions from soil and anthropogenicsources. However, variations in local soil composition fromcrustal averages can significantly blur this result. The EFs of�2 for K and Fe suggest that these elements have very largesoil contributions. For the remaining elements, the EFs varyfrom 140 to 13700. These elements are essentially ofanthropogenic origins. We also investigated variations inEFs over time. All elements except Pb, and Hg showed nodistinct variation with time at Mayville, or WhitefaceMountain. The EF for Fe, Hg, and Pb are shown in Figure 5.The EF for Hg, and Pb decreased dramatically from 1979 toabout 1987 and since then have decreased slowly. However,the EFs for Fe (Figure 5) have remained essentiallyunchanged. Since the primary source of atmospheric Pb inthe 1970s and early- to mid-1980s was leaded gasoline, it isreasonable to suggest that the ban on Pb additive in 1978sharply reduced its contribution to atmosphere. The EFdecrease for Hg has been comparable to that of Pb,

Table 1. Annual Means for Selected Trace Metals at Mayville and Whiteface Mountain, New Yorka

Year Kb Scb (pg/m3) Mnc Fed Znd Asd Sed Sbb Hgb Pbc

Whiteface Mountain1979 61.8 18.5 3.5 88.1 17.0 0.50 0.46 0.27 0.148 27.319801981 57.9 16.0 3.3 81.0 11.4 0.53 0.42 0.18 0.089 13.619821983 40.8 13.8 1.9 58.7 11.4 0.24 0.35 0.13 0.034 18.91984 47.0 16.6 2.3 70.1 8.7 0.36 0.41 0.37 0.063 13.41985 46.7 13.7 2.0 66.7 8.8 0.38 0.40 0.23 0.058 7.11986 40.9 13.5 2.1 68.1 11.3 0.33 0.38 0.20 0.042 6.51987 42.5 13.0 1.8 56.3 6.2 0.39 0.33 0.17 0.017 4.11988 49.9 14.9 2.5 72.3 8.8 0.39 0.44 0.16 0.023 7.11989 54.4 12.6 1.9 57.2 6.8 0.37 0.38 0.16 0.015 3.11990 43.3 12.2 1.7 52.0 10.0 0.27 0.38 0.16 0.018 2.51991 44.1 13.0 2.2 71.3 11.8 0.27 0.32 0.14 0.016 3.51992 24.1 8.0 1.1 40.0 6.8 0.25 0.24 0.14 0.011 3.11993 30.1 11.0 1.2 45.7 8.7 0.22 0.27 0.13 0.006 1.61994 28.9 12.9 1.2 54.1 9.6 0.25 0.37 0.18 0.009 2.01995 33.9 10.5 1.6 55.1 10.2 0.21 0.23 0.19 0.017 2.51996 33.1 10.0 1.3 51.7 10.1 0.29 0.28 0.21 0.007 1.51997 19.7 8.5 1.2 40.1 6.1 0.18 0.22 0.15 0.024 1.21998 29.9 9.8 1.8 55.0 6.2 0.18 0.30 0.14 0.002 1.11999 33.0 10.9 1.4 50.0 5.7 0.20 0.29 0.16 0.004 1.22000 0.9 28.1 3.5 0.19 0.25 0.11 1.42001 1.4 56.4 3.5 0.21 0.25 0.15 1.4

Mayville1984 154 59 8.9 224 28.9 1.18 1.52 0.53 0.063 30.91985 162 62 8.1 229 25.0 0.97 1.48 0.49 0.029 23.71986 162 62 8.6 234 27.4 1.17 1.69 0.61 0.035 22.21987 173 61 8.7 227 25.1 1.03 1.45 0.54 0.020 14.41988 160 62 9.1 236 23.7 1.26 1.57 0.58 0.039 10.91989 139 48 10.5 215 25.2 1.10 1.58 0.52 0.046 10.91990 128 51 9.5 214 25.5 0.96 1.66 0.52 0.034 9.11991 140 45 8.5 195 22.3 0.85 1.40 0.52 0.034 8.11992 102 33 6.6 140 17.7 0.72 1.18 0.45 0.033 6.01993 133 45 9.2 198 19.0 0.89 1.46 0.44 0.028 6.91994 120 38 7.1 164 18.0 0.83 1.31 0.44 0.011 9.71995 105 29 7.1 138 13.5 0.70 1.12 0.33 0.003 6.21996 110 34 6.8 142 14.3 0.86 1.29 0.40 0.014 10.71997 86 32 6.1 125 12.7 0.69 1.17 0.35 0.010 4.71998 159 67 10.6 224 15.6 0.81 1.46 0.47 0.015 5.11999 148 44 7.6 177 14.8 0.80 1.48 0.47 0.010 3.52000 6.1 141 10.8 0.82 1.50 0.38 4.72001 5.8 158 12.9 0.85 1.38 0.44 4.6

aUncertainties for annual means: K, ±5.0%; Sc, ±6.6%; Mn, ±5.0%; Fe, ±5.8%; Zn, ±7.0%; As, ±4.2%; Se, ±4.8%; Sb, ±4.4%; Hg, ±9.8%; Pb, ±8.4%.bINAA only.cICP-MS.dMean of ICP-MS and INAA determinations.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

6 of 12

D18305

although, the sources of the two elements are quite different.Fossil fuel combustion in power plants is considered one ofthe major sources of air pollution including fine particulatematter and Hgo [Hopke et al., 2003]. Electric powergeneration in the midwestern states is almost exclusivelyfrom coal combustion. Selenium is generally used as amarker element to trace coal combustion [e.g., Dutkiewicz etal., 1987, and references therein] but no decline in the EF ofSe is evident at either of our two sites. Coal usage hasactually increased and the steps taken to reduce SO2

emissions may not reduce trace elements like Se and Hg. Itis therefore unlikely that the decrease in Hg enrichment isrelated to changes in the pattern of coal combustion. Olmezet al. [1998] used air trajectories and factor analysis toevaluate sources of particulate Hg in samples collected atfive New York State sites between 1991 and 1993. Theyconcluded that the copper smelters in Canada were thepredominant source. They also indicated that elevated As/Seratios supported this conclusion. The EFs of As, however,do not significantly decrease at either site so the source ofthe decreasing Hg is unclear. The only other element toshow any temporal trend was Mn and in this case the EFs in1979 and 1981 were high but since then have remainedessentially constant. The trends at Mayville are quiteconsistent with those at Whiteface Mountain.3.2.2. Seasonal Trends[17] The quarterly data were obtained for 21 years at

Whiteface Mountain and for 18 years at Mayville, thusproviding a robust database with which we can investigateseasonal variations in the elemental concentrations and alsoto delineate changes, if any, as a function of time. Owing to

the fact that the concentrations of several elements havedecreased severalfold from 1979 to 2001, we have not usedsimple arithmetic means of the quarterly concentrations.Instead, we have used the deviations in the quarterlyconcentrations from the annual means to reflect seasonalchanges. Seasonal profiles at the two sites are plotted inFigure 6 for the crustal elements K, Sc, and Fe and for theanthropogenic elements Zn, As, Se, Pb, and particulate Hg.At Whiteface Mountain, the crustal elements (also includingMn, which is not shown) all show a very similar behaviors:about 40% lower than the annual mean in the first and thefourth quarters; and about 60% higher in the second quarterand �20% higher in the third quarter. At Mayville, the trendis similar but with magnitudes much reduced. For the first,and third quarters, the concentrations are within ±5% of theannual means for all four crustal elements. The second-quarter concentrations are 10 to 25% higher than the annualmeans and the fourth quarter concentrations are 10 to 25%lower. Since the Whiteface Mountain site is on a mountainsummit (1.5 km elevation), high winds and snowy con-ditions are prevalent during the first and fourth quarters andmuch of the summit is coated in rime ice. As the residencetime of particulate matter under these conditions is probablyvery short, this could depress the first and fourth quarterconcentrations enhancing the seasonal differences. A 60%reduction in crustal material during the first and fourthquarters based on these ‘‘mountain effects’’ could accountfor the differences in the seasonal pattern of the crustalelements at the two sites (Figure 6).Malm et al. [2004] havereported that the soil component at high-altitude sites in thenortheastern United States peaks during April, May, andJune. This is consistent with the Whiteface Mountain data.Mayville is a surface site. It still shows higher concentra-tions during the second quarter but not as strongly asWhiteface Mountain. Apparently, the crustal component istransported from distant sources above the mixing layer andas such would more strongly affect the sampling site atWhiteface Mountain, which is above the mixing layerduring much of the second quarter.[18] The anthropogenic elements in general exhibit little

seasonal variation (Figure 6). For example, at Mayville, Zn,As, and Pb had essentially the same concentrations for allfour quarters. At Whiteface Mountain, As concentrationswere within ±10% in all quarters, and Pb concentrationswere within ±20%, but Zn showed a 40% enhancement inthe third quarter. Selenium concentrations at Mayvilleshowed little seasonal variation, except for a 10% enhance-

Figure 5. Annual mean enrichment factors for Pb, Hg, andFe at Whiteface Mountain, New York. The lines are theresult of three-point running means.

Table 2. Enrichment Factors for Elements Relative to Sc at

Whiteface Mountain and Mayvillea

Mayville Whiteface Mountain

K 2.5 ± 0.3 2.3 ± 0.4Mn 4.2 ± 0.8 3.3 ± 0.5Fe 1.8 ± 0.2 2.1 ± 0.2Zn 140 ± 25 230 ± 52As 245 ± 45 290 ± 55Se 13740 ± 2590 11950 ± 1230Sb 1140 ± 190 1670 ± 370Hg 150 ± 85 450 ± 335Pb 390 ± 195 500 ± 325

aMean and ±1 standard deviation for enrichment factors; crustalabundances were taken from Mason [1966].

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

7 of 12

D18305

Figure 6. Quarterly mean concentrations for selected trace metals measured at Whiteface Mountain(solid diamonds) and Mayville (open squares), New York. Concentrations are normalized to the annualmeans to highlight seasonal trends. Solid lines are weighted multiple regression curves drawn through thedata points as a visual aid.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

8 of 12

D18305

ment in the third quarter. Particulate Hg at Mayville(Figure 6) behaved similarly to other anthropogenicelements, but at Whiteface Mountain it showed a 30%enhancement in third quarter and a 10% decrease in theother three quarters. We also investigated whether therewere any trends in seasonal variation over time, but wefound none.

3.3. Long-Term Trends

[19] Because of the strong seasonal trend for crustalelements at Whiteface Mountain, we show the Sc timeseries at this site both by quarter and on an annual basis,in Figure 7. A decreasing trend is evident in each quarter aswell as in the annual means. The lines are exponentialregressions. While the quarterly regression coefficients areonly marginally good for Sc (r2 between 0.28 and 0.51), allyielded trends between �2.6 and 3.9% per year, i.e., adecrease of 2.6 and 3.9% per year. For the annual meandata, r2 = 0.69 and the trend is �3.0% per year. Thus thedecreasing trend is consistent throughout the year. Alsoshown in Figure 7 is the trend in national PM10 emissionsextracted from an USEPA report [U.S. Environmental

Protection Agency, 2001]. There is also a steady decreasingtrend in the PM10 emissions, however, the decrease is only1.7% per year. Approximately 38% of the national PM10

emissions is from fuel combustion in motor vehicles, powergeneration, industrial facilities and space heating. A com-parable fraction is from industrial processing, and theremaining fraction from transportation. Unfortunately wedo not have the needed emissions data available for com-parison. Regression coefficients for Sc and other elements atMayville, and Whiteface Mountain, are given in Table 3.The values of r2 for Mayville are comparable to those atWhiteface Mountain. The Sc concentrations at Mayvilledecreased 4.7% per year, compared to 3% per year atWhiteface Mountain. The different rates of decrease, per-haps reflect the uncertainly of the measurements and fittingprocess.[20] For additional selected metals concentration varia-

tions with time are shown in Figure 8. As in Figure 7, thelines are exponential regression fits, and data from bothMayville and Whiteface Mountain are shown. The trends forall elements (not only the ones shown in Figures 7 and 8)were fitted with exponential curves, the fitting coefficientsare summarized in Table 3. Clearly, trace metal concentra-tions (Table 1) have been decreasing over time at both sites.The decreases in concentrations vary from 2.9 to 16%/year,although most of them cluster at between 2.9 and 5.5% peryear (Table 3). The exceptions are Pb and Hg, which havebeen decreasing at between 12% and 16% per year. Con-sidering all the possible variables that can impact aerosolconcentrations, it is somewhat surprising that the year-to-year variations of the annual means of all the elements aresystematic, although relatively subtle. The overall decreas-ing trend is seen in elements from diverse sources: SO4, Se,Sb, and As largely in PM2.5 from coal burning, K, Sc, Fe,and Zn in PM10 largely from the crust and from industrialprocesses; Pb and Hg in PM2.5 from transportation/fossilfuel combustion. The exponential curve fitting generallyyielded higher values of r2 at Whiteface Mountain than atMayville. Mayville is closer to the pollution sources than isWhiteface Mountain, particularly the large coal-burningutilities and iron and steel mills in the midwestern UnitedStates. Emissions from relatively local sources are less

Figure 7. Quarterly and annual mean scandium concen-trations at Whiteface Mountain, New York. Also shownare the trends in the national PM10 emissions [U.S.Environmental Protection Agency, 2001].

Table 3. Exponential Regression Coefficients and Exponents for

Trace Elements Analyzeda

Mayville Whiteface Mountain

r2 Exponent, %/year r2 Exponent, %/year

Crustal ElementsPotassium 0.82b �4.3 0.62 �4.0Scandium 0.68b �4.7 0.69 �3.0Manganese 0.60b �2.5 0.65 �4.4Iron 0.69b �3.4 0.52 �2.9

Anthropogenic ElementsZinc 0.88 �5.4 0.53 �4.4Arsenic 0.64 �3.2 0.73 �4.3Selenium 0.55c �2.5 0.68 �2.9Antimony 0.69c �3.6 0.27Mercury 0.50 �12 0.77 �16Lead 0.81 �10 0.90 �14

aAll results are for annual means.bSingle outlier removed.cRegression computed through 1997 only.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

9 of 12

D18305

dispersed, and hence are more susceptible to concentrationfluctuations due to meteorological conditions. In addition, asthe impact from long distant sources decreases, the impactfrom local sources (even if small) can increase in a relativesense. We suspect that the latter phenomenon is the reasonfor the recent increases in Se, As and Sb concentrations atMayville.[21] Figure 8 shows the concentration variation with time

for the important marker elements; Se, As and Sb which areused in receptor modeling [e.g., Rahn and Lowenthal, 1985;Dutkiewicz et al., 1987; Olmez et al., 1998]. A compilation

and discussion of Se measurements from around the north-eastern United States for the early to mid-1980s are givenby Dutkiewicz and Husain [1988]. Concentrations of Se atWhiteface Mountain (Figure 8) varied from a high of 0.46in 1979 to a low of 0.22 ng m�3 in 1997. The concentrationin 2001 was 0.25 ng m�3. Over the whole period, Sedecreased by 46%. The exponential curve has r2 = 0.65which is a modest fit, at best. There is a flattening in theslope of the concentration curve between 1982 and the early1990s, reminiscent of the sulfate and SO2 emission curves(Figure 1). Annual mean PM10 Se concentrations for1993 atEvans Center, New York [Sweet et al., 1998] are also shownin Figure 8. The site is located along the shore of Lake Erievery near Mayville. The Se concentration at Evans Centerwas 1.7 ng m�3, in excellent agreement with the value of1.5 ng m�3 observed at Mayville. In 1998, however, the Seconcentration at Mayville increased to around 1.5 ng m�3

where it remained through 2001. As the Se increases after1997 at Mayville are not echoed at Whiteface Mountain, theMayville Se increases appear to represent a localizedphenomenon. The exponential curve for Mayville inFigure 8 is a poor fit to the Se data. Limiting the fittingto the years through 1997 improves the r2 to 0.55, andyielded a decreasing rate of 2.5% per year.[22] The As concentration at Whiteface Mountain was the

highest in 1981, 0.53 ng m�3, and the lowest in 1997 and1998, 0.18 ng m�3 (Figure 8). Such a change represents a66% decrease. In 1999–2001 the concentration remainedaround 0.20 ng m�3. The exponential regression yieldedr2 = 0.73 and a rate of decrease of 4.3% per year,comparable to the values for crustal elements. Arsenicmeasured at Evans Center was 0.9 ng m�3, in excellentagreement to the measurements at Mayville. For Mayville,the exponential curve was not a very good fit to the data, asr2 = 0.63. Nevertheless, the trend is very similar to that atWhiteface Mountain, �3.2% per year.[23] The highest Sb concentration at Whiteface Mountain

(Figure 8) was 0.37 ng m�3, recorded in 1984, and thelowest was 0.11 ng m�3, measured in 2000. Since 1992 wehave occasionally collected 2 to 6 hour aerosol samples atthe summit of Whiteface Mountain over short stretches oftime during July and August. These samples were analyzedfor Se, As, and Sb by INAA [Burkhard et al., 1995]. Weobserved very high Sb concentrations in a few samples thatcould not be related to any other element measured. Duringsome of these campaigns a second set of samples werecollected at a nearby site at an elevation of 600 m (Lodge).When antimony spikes were recorded at the summit therewere none at the Lodge site. Thus some intermittentcontamination of antimony is suspected at the summit.Therefore the Sb results at Whiteface Mountain are reportedfor completeness, we will limit further discussion to theresults from Mayville. At Mayville the highest concentra-tion, 0.61, was measured in 1986, and the lowest in 1995,0.33 and similarly, 0.35 ng m�3 in 1997. At Mayville Sbwas 0.44 ng m�3 in 2001. The exponential curve forMayville (Figure 8) has r2 = 0.48 so it is not an adequatemodel for the data. However, limiting the fitting range to1997 improves r2 to 0.81 and yields a trend of �3.6% peryear.[24] The variation of Hg concentrations with time are also

shown in Figure 8. The highest Hg concentrations at

Figure 8. Annual mean concentrations of selected tracemetals measured at Mayville (open squares) and WhitefaceMountain (solid circles), New York. The dashed and solidlines are exponential regression fits to the data. Regressioncoefficients and rate constants are summarized in Table 3.Also shown, as indicated in the key, are aerosol measure-ments from sites near Mayville and Whiteface Mountain.Data from Evans Center are from Sweet et al. [1998], whiledata from Westfield, Moss Lake, and Willsboro are fromOlmez et al. [1998].

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

10 of 12

D18305

Whiteface Mountain, 0.15 ng m�3, occurred in 1979, andconcentrations have since steadily decreased. The mostrecent measurements, 0.004 ng m�3, were for 1999. Keeleret al. [1995] found that particulate Hg at rural areas of theGreat Lakes Region and Vermont, averaged between 0.009and 0.022 ng m�3, while in downtown Detroit it averaged0.094 ng m�3, with 60–100% in the PM2.5 fraction.Olmez etal. [1998] reported aerosol Hg measurements from 5 sites inupstate New York between December 1991 and September1993. Arithmetic means, minus outliers, varied from 0.036 to0.057 ng m�3. The data from the three sites nearest White-face Mountain and Mayville are shown in Figure 8. West-field, as mentioned earlier, is very near Mayville while MossLake and Willsboro are near Whiteface Mountain. Concen-trations measured at these sites are very similar to those atWhiteface Mountain and Mayville. At Mayville the earliestmeasurement was also the highest, 0.063 ng m�3 in 1984; by1999 concentrations had decreased to 0.010 ng m�3. Theseconcentrations are in the same range as those reported byKeeler et al. [1995] and Olmez et al. [1998]. The exponentialregression fit at Mayville was marginal, as r2 was only0.5, compared to 0.77 at Whiteface Mountain. The rates ofconcentration decrease were similar at 12% and 16% peryear, respectively. A discussion of likely source regionsimpacting these sites can be found in Olmez et al. [1998].[25] Concentrations of particulate Hg relative to gaseous,

may depend upon the distance from the emission point.Xiao et al. [1991] measured 15.3% of the total atmosphericHg as particulate at a station close to an industrial area whilethe average was 4.5% for all other sampling sites. Hopke etal. [2003] recently reported the measurements of Hgo atStockton, 14 km northeast of Mayville, and Potsdam, 75 kmnorthwest of Whiteface Mountain. Mean Hgo values forStockton were 1.23 and 1.62 ng m�3, respectively, for 2000and 2001. Hgo concentrations at Potsdam were 2.43 and1.14 ng m�3, respectively, for 2000 and 2001. Thus in 2000the highest concentrations were at Potsdam while in 2001the highest concentrations were at Stockton. The authorsattributed this to differences in the meteorological condi-tions which led to different source regions impacting their

sampling sites. Hopke et al.’s [2003] data can be used toestimate the particulate Hg portion of the total Hg. Althoughwe did not measure Hg during 2001 and 2000, the data inFigure 8 shows little variation in particulate Hg during1995–1999 at either of our two sites. During 1999 partic-ulate Hg averaged 0.004 and 0.01 ng m�3, respectively, atWhiteface Mountain and Mayville. Thus based on averageHgo concentrations of 1.4 and 1.79 ng m�3, respectively, forStockton and Potsdam, we estimate that particulate phaseHg concentrations would constitute only a few percent ofthe Hgo.[26] The Pb concentrations with time are shown in

Figure 9. Concentrations at Whiteface Mountain variedfrom a high of 27.3 ng m�3 in 1979 to 1.1 ng m�3 in1998. Concentrations have subsequently risen slightly, to1.4 ng m�3 in 2001. At Mayville the highest concentrations,30.9 ng m�3, was measured in 1984, and the lowest,3.5 ng m�3, in 1999. Like Whiteface Mountain, concen-trations are up slightly in 2001 to 4.6 ng m�3. On averagePb concentrations at Whiteface Mountain are lower than atMayville, by a factor of 3. The pattern for Pb (Figure 9)clearly depicts a distinct downward trend and the exponen-tial regression summarized in Table 3 is a good fit to thedata at both sites. The data shows a decrease of 14% year�1

at Whiteface Mountain and 10% year�1 at Mayville.Between 1981 and 2000 national Pb emissions decreasedby 94% while over the same period Pb concentrationsat Whiteface Mountain decreased by 90%. Overall, Pbconcentrations at Whiteface Mountain and Mayville are areasonably good fit to the National emission trend shown(Figure 9). Lead measured at Evans Center during 1993–1994 averaged 7.9 ng m�3, in excellent agreement withconcentrations measured at Mayville.

4. Summary

[27] Concentrations of K, Sc, Mn, Fe, Zn, As, Se, Sb, Hgand Pb were determined in quarterly composites of dailyaerosol samples collected from 1984 to 2001 at Mayville,New York and 530 km downwind at Whiteface Mountain(1.5 km altitude) from 1979 to 2001. SO4 concentrationswere determined in individual daily samples. In addition,continuous hourly SO2 data are available from 1988 onwardat Whiteface Mountain and from 1992 to the present at asite 15 km from Mayville. The SO4 concentrations atWhiteface Mountain decreased by 18% from 1979 to1981 but then remained fairly flat through 1991, anddecreased by �14% compared to 1981–1991 in 1992–1994. A sharp decrease of 21% occurred from 1994 to 1995.Since 1995 the concentrations have remained essentially thesame. The variations at Mayville have been similar. Thetrends in SO4 concentrations have mimicked the variationsin regional SO2 emissions upwind of and contiguous withNew York State (PA, OH, IN, IL, WI, KY, WV and ONT).In fact, the SO4 concentrations were linearly related to theSO2 emissions for the entire period. At Mayville, the ratioof SO4 concentrations to the SO2 emissions was 0.200 ±0.018 mg m�3 per kmt of SO2 per day. At WhitefaceMountain, this ratio was 0.100 ± 0.009 mg m�3 per kmtof SO2 per day from 1979 to 1992, but it has slightlydeclined to 0.082 ± 0.009 from 1997 to 2002. The datasuggests that the SO4 may now be decreasing �20% faster

Figure 9. National emissions trend for lead, normalized toannual mean lead concentrations at Mayville (opentriangles) and Whiteface Mountain (open squares), NewYork. Lines are three-point running means, while resultsfrom exponential regression fits are given in Table 3.

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

11 of 12

D18305

at Whiteface Mountain than the SO2 emissions relative to1979–1996 period. This may be due to the decrease in theabsolute amount of the emissions or due to spatial changesin regional emissions. Total S (sum of S present as SO2 plusSO4) concentrations behaved similar to the SO4 levels. Theabove observations suggest that any reductions in the SO2

emissions would result in a proportional decrease in atmo-spheric S burden, across New York State, and possiblyacross northeastern US.[28] The quarterly trace element concentrations showed

higher levels in the second and third quarters, relative to theannual mean, and lower in the first and fourth quarters. Withvarying degree this held for all elements except As, Se andparticulate Hg at Whiteface Mountain. The seasonal trendswere, in general, weak at Mayville relative to WhitefaceMountain. The trace element concentrations showed unmis-takable decrease in concentrations over time. Apparently, theregulations restricting emissions of SO2 and particulatematter have also resulted in a decrease of atmosphericburden of trace elements. The largest decreases were ob-served for Hg (16%/year at Whiteface Mountain and 10%/year at Mayville), and Pb (14%/year at Whiteface Mountainand 10%/year at Mayville). The remaining elements (exceptSb) including the crustal elements, K, Mn, Sc and Fe,showed decreases of 3 to 5%/year. Trends for Sb at White-face Mountain and Mn at Mayville could not be accuratelydiscerned, apparently due to some nearby emissions.

[29] Acknowledgment. The authors are grateful to Steve Johnson,Robert Lincoln and Don Swingle of the Chautauqua County HealthDepartment for collecting samples at Mayville and the AtmosphereSciences Research Center, SUNYAlbany for the use of sampling facilitiesand Douglas Wolfe and Paul Casson for sample collection at WhitefaceMountain.

ReferencesBurkhard, E. G., V. A. Dutkiewicz, B. M. Ghauri, and L. Husain (1995), Amulti-element tracer technique for determination of in situ oxidation ofSO2 in cloud droplets, J. Geophys. Res., 100, 26,051–26,059.

Dutkiewicz, V. A., and L. Husain (1988), Spatial patterns of non-urban Seconcentrations in the northeastern US and its pollution source implica-tions, Atmos. Environ., 22, 2223–2228.

Dutkiewicz, V. A., P. P. Parekh, and L. Husain (1987), An evaluation ofregional signatures relevant to the northeastern United States, Atmos.Environ., 21, 1033–1044.

Dutkiewicz, V. A., M. Das, and L. Husain (2000), The relationship betweenregional SO2 emissions and downwind aerosol sulfate concentrations inthe northeastern US, Atmos. Environ., 34, 1821–1832.

Gone, J., I. Olmez, and M. R. Ames (2000), Size distribution and possiblesources of trace elements in submicron atmospheric particulate material,J. Radioanal. Nucl. Chem., 244, 133–139.

Hicks, B. B., R. S. Artz, T. P. Meyers, and R. P. Hosker Jr. (2002), Trends ineastern U.S. sulfur air quality from the Atmospheric Integrated Monitor-ing Network, J. Geophys. Res., 107(D12), 4143, doi:10.1029/2000JD000165.

Hopke, P. K., L. W. Liu, Y. Han, S. Yi, T. M. Holsen, S. Cybart, andM. Milligan (2003), Measured summertime concentrations of particulatecomponents Hgo, and speciated polycyclic aromatic hydrocarbons at ruralsites, Environ. Pollut., 123, 413–425.

Husain, L., and P. J. Samson (1979), Long range transport of trace ele-ments, J. Geophys. Res., 84, 1237–1240.

Husain, L., J. S. Webber, E. Canelli, V. A. Dutkiewicz, and J. A. Halstead(1984), Mn/V ratio as a tracer of aerosol sulfate transport, Atmos. Environ.,18, 1059–1071.

Husain, L., V. A. Dutkiewicz, and M. Das (1998), Evidence for decrease inatmospheric sulfur burden in the eastern United States caused by reduc-tion in SO2 emissions, Geophys. Res. Lett., 25, 967–970.

Keeler, G., G. Glinsorn, and N. Pirrone (1995), Particulate mercury in theatmosphere: Its signature, transport, transformation and sources, WaterAir Soil Pollut., 80, 159–168.

Malm, W. C., B. A. Schichtel, R. B. Adams, and K. A. Gebhart (2002), A10-year spatial and temporal trend of sulfate across the United States,J. Geophys. Res., 107(D22), 4627, doi:10.1029/2002JD002107.

Malm, W. C., B. A. Schichtel, M. L. Pitchford, L. L. Ashbaugh, and R. A.Eldred (2004), Spatial and monthly trends in speciated fine particle con-centration in the United States, J. Geophys. Res., 109, D03306,doi:10.1029/2003JD003739.

Mason, B. (1966), Principles of Geochemistry, 3rd ed., John Wiley, Hobo-ken, N. J.

Olmez, I., M. R. Ames, and G. Gullu (1998), Canadian and US sourcesimpacting the mercury levels in fine atmospheric particulate materialacross New York State, Environ. Sci. Technol., 32, 3048–3054.

Rahn, K. A., and D. H. Lowenthal (1984), Elemental tracers of distantregional pollution aerosols, Science, 223, 132–139.

Rahn, K. A., and D. H. Lowenthal (1985), Pollution aerosols in the north-east: Northeastern-midwestern contributions, Science, 228, 275–284.

Swami, K., C. D. Judd, J. Orsini, K. X. Yang, and L. Husain (2001),Microwave assisted digestion of atmospheric aerosol samples followedby inductively coupled plasma mass spectroscopy determination of traceelements, Fresenius J. Anal. Chem., 369, 63–70.

Sweet, C. W., A. Weiss, and S. J. Vermette (1998), Atmospheric depositionof trace metals at three sites near the Great Lakes, Water Air Soil Pollut.,103, 423–439.

U.S. Environmental Protection Agency (2001), Latest findings on nationalair quality: 2001 status and trends, report, Off. of Air Qual. and Stand.,Washington, D. C.

Xiao, Z. F., J. Munthe, and O. Lindqvist (1991), Sampling and determina-tion of gaseous and particulate mercury in the atmosphere using gold-coated denuders, Water Air Soil Pollut., 56, 141–151.

Yang, K. X., K. Swami, and L. Husain (2002), Determination of tracemetals in atmospheric aerosols with heavy matrix of cellulose by micro-wave digestion-inductively coupled plasma mass spectroscopy, Spectro-chim. Acta, Part B, 57, 73–84.

�����������������������V. A. Dutkiewicz, L. Husain, A. R. Khan, P. P. Parekh, K. Swami, and

K. Yang, Wadsworth Center, New York State Department of Health,Albany, NY 12201-0509, USA. ([email protected]; [email protected]; [email protected]; [email protected]; [email protected])

D18305 HUSAIN ET AL.: TRENDS IN ATMOSPHERIC SULFATE AND METALS

12 of 12

D18305