nitrogen and sulfur species in antarctic aerosols at ...dust.ess.uci.edu/ppr/ppr_spl93.pdf ·...

Journal of Atmospheric Chemistry 17: 95-122, 1993. 9 5 © 1993 Kluwer Academic Publishers. Printed in the Netherlands.

Nitrogen and Sulfur Species in Antarctic Aerosols at Mawson, Palmer Station, and Marsh (King George Island)

D. L. S A V O I E , J. M. P R O S P E R O , R. J. L A R S E N 1, H U A N G , F., M. A. I Z A G U I R R E , H U A N G , T., T. H. S N O W D O N , L. C U S T A L S , and C. G. S A N D E R S O N 1 Division of Marine and Atmospheric Chemistry, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149-1098, U.S.A. 1Environmental Measurements Laboratory, Department of Energy, New York, NY 10014, U.S.A.

(Received: 21 February 1992; in final form: 21 January 1993)

Abstract. High volume bulk aerosol samples were collected continuously at three Antarctic sites: Mawson (67.60 ° S, 62.50 ° E) from 20 February 1987 to 6 January 1992; Palmer Station (64.77 ° S, 64.06 ° W) from 3 April 1990 to 15 June 1991; and Marsh (62.18 ° S, 58.30 ° W) from 28 March 1990, to 1 May 1991. All samples were analyzed for Na +, SO 2-, NO3, methanesulfonate (MSA), NH~, 21°pb, and 7Be. At Mawson for which we have a multiple year data set, the annual mean concentration of each species sometimes vary significantly from one year to the next: Na +, 68-151 ng m-3; NO?, 25-30 ng m 3; nss SO]-, 81-97 ng m-3; MSA, 19-28 ng m 3; NH], 16-21 ng m-3; 21°pb, 0.75-0:86 fCi m -3. Results from multiple variable regression of non-sea-salt (nss) SO42- with MSA and NO3 as the independent variables indicates that, at Mawson, the nss S O ] / M S A ratio resulting from the oxidation of dimethylsulfide (DMS) is 2.80 ___ 0.13, about 13% lower than our earlier estimate (3.22) that was based on 2.5 years of data. A similar analysis indicates that the ratio at Palmer is about 40% lower, 1.71 _+ 0.10, and more comparable to previous results over the southern oceans. These results when combined with previously published data suggest that the differences in the ratio may reflect a more rapid loss of MSA relative to nss SO4 z- during transport over Antarctica from the oceanic source region. The mean 21°pb concentrations at Palmer and Marsh and the mean NO3 concentration at Palmer are about a factor of two lower than those at Mawson. The 21°Pb distributions are consistent with a 21°pb minimum in the marine boundary layer in the region of 40°-60 ° S. These features and the similar seasonalities of NO 3 and 21°pb at Mawson support the conclusion that the primary source regions for NO 3 are continental. In contrast, the mean concentrations of MSA, nss SO] , and NH2 at Palmer are all higher than those at Mawson: MSA by a factor of 2; nss SO42- by 10%; and NH~ by more than 50%. However, the factor differences exhibit substantial seasonal variability; the largest differences generally occur during the austral summer when the concentrations of most of the species are highest. NH~/(nss SO 2- + MSA) equivalent ratios indicate that NH 3 neutralizes about 60% of the sulfur acids during December at both Mawson and Palmer, but only about 30% at Mawson during February and March.

Key words. Antarctica, Palmer, Marsh, Mawson, aerosol particles, biogeochemical cycles, sulfate, nitrate, methanesulfonate, lead-210, beryllium-7, sea-salt, ammonium.

1. Introduction

Understanding the atmospheric chemistry of the Antarctic is crucial for several reasons. Because this region is far removed from major pollution sources, the

96 D.L. SAVOIE ET AL.

atmospheric cycles of most major components are minimally impacted by products of man's activities. Long-term measurements at numerous locations in the region are required along with analyses of recent snow chemistry to substantiate the assumption that the vertical profiles of species in snow and ice cores provide a reasonably unam- biguous record of the atmosphere's historical composition and climate.

The objective of our ongoing study is to further our understanding of the temporal variations and the probable sources of the major sulfur and nitrogen species in aerosols over Antarctica and the southern oceans. To aid in the assessment of the sources, we also measure the concentrations of two natural radioactive tracers, 21°pb and 7Be. Beryllium-7 (tl/2 = 53.6 d) is a tracer for transport from the lower stratosphere and upper troposphere where it is produced by the cosmic ray- induced spallation of nitrogen and oxygen. Lead-210 is a decay product of 222Rn (tl/2 = 3.8 d) for which the overwhelmingly dominant source is emission from soils. Consequently, 21°pb se rves as a tracer for air masses that have recently been in con- vective contact with continental land masses.

In two recent reports (Prospero et al., 1991; Savoie et al., 1992), we presented and discussed the results that we had obtained from the analyses of aerosol samples collected at Mawson, Antarctica (67.60 ° S, 62.50 ° E; Figure 1) from February 1987 through October 1989. In this report, we build on the results from our previous study by examining the additional nss SO]-, MSA, NO~, NH~, 21°Pb, and 7Be data that were acquired at Mawson through December 1991. This larger data set provides an opportunity to assess the interannual variations in the mean concentrations as well as in the seasonal cycles. The larger number of samples also significantly reduces the statistical uncertainties in the regression coefficients for esti- mating the nss SO]- concentrations from those of MSA and NO~ and in the mean NH~: nss SO]- ratio. To enhance our understanding of the spatial distributions of the concentrations, the Mawson data are compared to those that we have recently obtained at two stations in the vicinity of the Antarctic peninsula: Palmer Station (64.77 ° S, 64.05 ° W); and Marsh, King George Island (62.18 ° S, 58.30 ° W). Sub- stantial local pollution at the Marsh site was evidenced by frequently extremely high concentrations and erratic variations in NO~ and nss SO 2- and by intense blackening of nearly all of the samples by soot carbon. This feature limits the use- fulness of the Marsh data and, consequently, discussion of data from that station is confined to Na +, MSA, 21°Pb, and 7Be.

2. Experimental Procedures

For sample collection, air was continuously drawn through 20 x 25 cm Microdon filters at a nominal flow rate of 50 m 3 h -1. The actual flow rates were determined from the pressure drops across a calibrated orifice system attached to the pump outlets. Weekly sampling periods yield sampled air volumes of about 8500 m 3. Results from tests performed for the U.S. Department of Energy (New York) by the Centre in Aerosol and Technology, Laurentian University (Sudbury, Ontario,

NITROGEN AND SULFUR IN ANTARCTIC AEROSOLS 97

9 0 ° W

Fig. 1.

I . . .

) . "

. .

• . °

0 °

. . . .

• . • • . •

" . . , . " . - • .

• . . . . . . . . 6 o - . . . . : . . " , . .

• " . G ' e o r 9 vo'n Neumajer. • , , " , • ' " " " . . . . . . . . . . . . - .

m

- MAWSON . p A L M E R i

• . , . • .

. . . [ . . . . : . . . . . . . . . :

• . . - • •

Sout~Pole

'- ,•"

C r o z e t O . . , . . . ,

v q . . ' .

~ ..kerg,,el.ene" "".~ "" • •"." " A m £ " l : e r d a r h l l

e '~ l • ~ . . . . - . . . .

. . . . i . . . . j ., - . . . . ... . . . . : . .

..."... ' " . i . . " . . .. '":i...:..".. . : . . " / . . / " . . . . . . ~ i i i ."." .'.i ." "" ....... .;----~umb6t d'Urv'me.......].~

: : 7

180 °

Map of the Antarctic and southern ocean showing the locations of some of the major stations.

9 0 ° E

Canada), indicate that, the Microdon filters have a collection efficiency of essentially 100% for test particles larger than about 0.2 ~m diameter and about 98% for those of about 0.1 ~m diameter. Blanks (taken after every third sample) are handled in the same way as samples except that they have had no air drawn through them.

The chemical analyses are performed at the University of Miami. For analysis of the water-soluble species, a one-eighth section of each filter is first wetted with 1 mL of ethanol and then extracted with 30 mL of Milli-Q water (18 Mg2 cm) in three separate aliquots of 10 mL each• NO~, SO 2-, CI-, and MSA concentrations in the extract are determined to within + 5% by ion chromatography• Na + is measured to within + 2% by flame atomic absorption and NH~ to within + 5% by automated colorimetry. Non-sea-salt (nss) SO]- is the difference between total SO 2- and sea- salt SO]-, the latter being calculated as total Na + times 0•2517 (the SO ]- :Na + mass ratio in bulk seawater)• As discussed elsewhere (Savoie et al., 1989a), the NO~ values reported here are for total inorganic NO~, i.e. particulate NO? plus gaseous H N O 3.


Although the consistency of the earlier NH] concentrations from sample-to- sample and blank-to-blank provided substantial evidence that NH~ contamination was not a problem at Mawson, we performed a more rigorous check during the period from February 1990 through January 1991. We prepared blanks by impreg- nating filter sections with sea-salt and nitric and sulfuric acids in a proportion that approximated that which we had actually observed in the Mawson samples during our first two years of study. One group of blanks was sent to Mawson to be exposed to laboratory air while the actual filter samples were being manipulated; another group was kept sealed and stored in our laboratory in Miami as controls. Analyses of these sections revealed that there was no significant difference between the ammonium concentrations in the two groups nor between the special blanks and the regular (untreated) filter blanks that are normally processed. These results provide confirming evidence that the ammonium concentrations in the samples are not significantly affected by contamination in the laboratory or in storage and ship- ment.

Although 7Be and 21°pb are produced in the gas phase, they are rapidly scav- enged by particulate material in the atmosphere and, as a consequence, are effi- ciently collected on the filters that are used in this study. The radionuclides are measured at the Department of Energy using a Ge(Li) well-type v-ray detector. Because of the long storage period that is forced by the winter-over at Mawson, there is often a large decrease in the concentration of 7Be. In many of the middle to late summer samples, the activity is so low as to be unmeasurable.

3. Mean Concentrations and Temporal Variations at Mawson

3.1. Interannual Variations

All of the species exhibit significant interannual variability (Figures 2-4, Table I). Sodium displays the most dramatic variability with a 1990-1991 mean that is more

Table I. Mean concentrations of various species during one year and multi-year periods at Mawson, Antarctica. Values shown are for the periods from 20 February of the first year to 20 February of the second

Period NO~ Na + NSS SO4 z- MSA NH~ 21°pb (ng m -3) (ng m -3) (ng m -3) (ng m -3) (ng m -3) (fCi m -3)

1987-1989 27.0 (17.4) 84.1 (64.0) 90.6 (88.3) 19.9 (21.6) 21.1 (22.0) 0.85 (0.40)

1987-1988 24.8 (16.5) 68.3 (50.0) 80.9 (87.0) 19.4 (25.1) 21.1 (22.0) 0.84 (0.35) 1988-1989 28.9 (18.4) 99.8 (73.2) 97.8 (88.5) 19.9 (17.7) 20.4 (21.7) 0.86 (0.45) 1989-1990 26.8 (17.3) 87.3 (98.0) 96.3 (97.7) 27.8 (34.6) 17.1 (21.0) 0.75 (0.37) 1990-1991 30.1 (19.3) 151.0 (142.0) 84.5 (92.6) 24.1 (24.7) 16.0 (19.2) 0.80 (0.41)

1987-1991 27.7 (17.9) 102.0 (101.0) 89.9 (91.2) 22.9 (26.3) 18.5 (20.9) 0.81 (0.39)

Values in parentheses are the sample standard deviations.


180 160 4-" 1,m

I E 12o

1 oo 13n ~o ~C 60, GO :~ 40

20, 0

400,

I 350, E 300.

250. i"i 200,

150. 1oo.

~z 50- 0

I O0 - 90Z 80~ 70: 60 5oi

- b 3~2 - r 20' -7 10:

O'

350

---,-. 300 re3

1 250 E 200 o ~. 15o

1~ lOO r--,.

20

0 ~

"1"9961 '19"9i

" . i f f B ~ . ~ . .

70 60

30 20 1

do" ¢e'Ma'Ap'tda'du "du'au'Se'0c'So'De'

~ 0 . . . . . 25o 2oo I ¢

N

2 °1 J Y'\l IOO-I , , I , 11 ~ i

50~

da Fe Mo Ap Ma du du Au Se Oc No De

Fig. 2. The graphs on the left show the actual time series of the weekly-average concentrations of NO?, methanesulfonate (MSA), non-sea-salt (nss) SOl-, NH], and 7Be at Mawson. The graphs on the right illustrate the composited seasonal, cycles, showing the monthly means and their 95% confidence intervals.

than double that for 1987-1988. For the other constituents, the ratio of the highest to the lowest annual means range from 1.12 for nitrate to 1.43 for MSA. For all species, the interannual changes are random, i.e., there are no consistent increasing or decreasing trends. Nor are the directions of the changes consistent from one species to another. The years with the highest concentrations are the final two for MSA, the first two for NH] and zl°Pb, and the middle two for nss SO4 a-. Although NO~ and Na ÷ do exhibit similar directions in their interannual changes, this feature is probably fortuitous since the species are not otherwise significantly correlated.

For many of the species, the means for the full four-year (1987-1991) period differ significantly (95% confidence level) from those for the two-year (1987- 1989) period. The four-year MSA, NH~, and Na + means are 15% higher, 12%


.---. 802 '7 70'

E 60 5o

"-" 40- I ~ 30". 0 z 20-

102

0" "'1"987 . . . . . . 1" 9"88 . . . . . . 1" 989 . . . . . . 1"9"96 . . . . . . i 9"9 i' '"

~" 1.8 I E 1.6

1.4 1.2

,~, 1.0 0.8 0.6 o

0.2 0.0 -- " i 9 8 7 1 i 9 8 8 I ~ffSbl i9~bI ibgi

50

4.0

30

20

10

01 . . . . . da'Fe'Ma Ap Ma'du 'du "Au'Se'Oc'No'De

1 . 2 ~ ] ~ @ @ ~ 1.0

0.8

0.6

0.4

0.2]

0.01 . . . . . . . . . . . . . . . . . . . da Fe Ida Ap Ma du du Au Se Oc No De

Fig. 3. The graphs on the left show the actual time series of the weekly-average concentrations of NO 3 and 21°pb at Mawson. The graphs on the right illustrate the composited seasonal cycles, showing the monthly means and their 95% confidence intervals.

700

600

500

400

300 v

I- o 200 "7 100'

0 i9i~7 f i g a s I -19891 i 996f i 9 9 i

250,

0 | . " . . - da'Fe'Ma'Ap Ma'du'du't£u'Se'Oc'No'[56

Fig. 4. The graph on the left show the actual time series of the weekly-average concentrations of Na + at Mawson. The graph on the right illustrates the composited seasonal cycle, showing the monthly means and their 95% confidence intervals.

lower, and 21% higher, respectively, than their 1987-1989 averages that we presented previously (Savoie et al., 1992). In contrast, the two means for NO3 and nss SO42- are essentially indistinguishable, differing by only 1-2%. The full four-year mean for 21°pb is only about 5% lower than that for the original two-year period. Because of the limited 7Be data for the summer periods at Mawson, we have not attempted to assess the interannual variation for that constituent. However, based on the average of the composited monthly means, we calculate an overall annual mean of about 140 fCi m -3, about 14% lower than our previous estimate of 162 (Savoie et al., 1992).

3.2. Seasonal Cycles

Despite the interannual variations in the means, the concentrations of each of the constituents at Mawson continue to exhibit reasonably consistent patterns in their


seasonal cycles (Figures 2-4). This consistency is reflected in the relatively small confidence intervals for most of the composited monthly means.

The concentrations of MSA, nss SO 2-, and NH~ all peak in the austral summer (Figure 2) at levels that are more than an order of magnitude greater than those during the winter. At their peaks during December and January, the monthly mean concentrations of these species are 57 -60 ,256 -272 , and 52-63 ng m 3, respectively, compared to means of 2, 10-11, and about 3 ng m -3 at their minima during June and July. The timing of these seasonal cycles is essentially identical to that of dimethylsulfide (DMS) in the air and in the surface ocean water of the Antarctic region (Ayers et al., 1991; Gibson et al., 1988, 1990; Nguyen et al., 1990). Hence, it is very likely that the oxidation of this sulfur gas which originates from biological activity in the ocean is the primary source of both MSA and nss SO]-. The source of NH~ is still unclear. However, the fact that its seasonal cycle is so similar to that of the sulfur species indicates that its ultimate source may also be linked to biological processes.

The seasonal cycle of 7Be is unique among the constituents that we measure, exhibiting a maximum in March and being nearly invariant from June through December (Figure 2). The physical reasons for this type of cycle are not clear, and a research effort on this topic is clearly in order. The results would suggest that the maximum impact of constituents from the lower stratosphere and upper troposphere occur during the late austral summer.

In contrast, the more complex nitrate cycle (Figure 3) exhibits its minimum (10- 12 ng m -3) from April through June, a relative maximum of 32 ng m -3 during August, and its annual maximum of 58 ng m -3 in November. Although we must concede that there is still a rather significant uncertainty in the composited monthly mean concentrations of 21°Pb (primarily a consequence of the relatively large uncertainties in the individual measurements), it is the species whose seasonal cycle most closely resembles that of NO 3. Moreover, the NO~ seasonal cycle differs substantially from those of MSA and 7Be. These observations lead us to conclude, as we have previously (Savoie et al., 1992), that the major source of the NO~ is continental as opposed to marine biogenic or stratospheric.

The seasonal cycle of sodium is less well defined (Figure 4). However, the maximum monthly means occur during the winter as do the highest peaks in the individual weekly values. This feature has been attributed to the winter storm activity over the southern oceans. The sporadic nature of both the occurrence of these storms and their penetration to the interior of Antarctica contribute to the highly irregular character of the Na + time series over most of the continent.

3.3. Mid-December 1988 Through Mid-February 1989

Although the seasonal cycles for each of the species are very consistent, the concentrations from mid-December 1988 through mid-February 1989 are unusual in several respects. In particular, the NO~ concentrations are significantly higher than


those during the same periods of the other years. From 16 December 1988 to 6 January 1989, the mean NO? concentration is 63 ng m -3 (weekly values range from 55.3 to 72.5 ng m -3) compared to means of 21.6 during 1987-1988, 32.6 during 1989-1990, and 40.9 during 1990-1991 (when the weekly values ranged from 19.5 to 47.2 ng m-3). The mean concentration of 56 ng m -3 from 20 January to 18 February 1989, is more than double the analogous mean during any other year. This period is even more exceptional because the means in all of the other years are very similar to one another: 27.7 in 1988; 22.5 in 1990; and 23.7 in 1991. The NH~ concentrations are also exceptionally high from 20 January to 18 February 1989 with a mean (52.9 ng m -3) that is 1.8 to 2.7 times higher than those of the other years (28.8 in 1988, 25.4 in 1990, and 19.7 in 1991). However, the NH~ concentrations from 16 December 1988 to 6 January 1989 are not particularly remarkable when compared to other years.

In contrast to those of NO~ and NH~, the concentrations of MSA from mid- December 1988 through mid-February 1989 are lower than those measured during any other comparable summertime period. In fact, the mean for this period (36.5 ng m -3) is 42% lower than that for 1987-1988 (62.8 ng m-3), 68% lower than that for 1989-1990 (87.6 ng m-3), and 26% lower than that for 1990-1991 (49.6 ng m -3)

When compared with NO~, NH], and MSA, the mean nss SOl- concentrations are unique in that they exhibit very little interannual variation. During our period of study, the nss SO ]. means from mid-December through mid-February change by a maximum of 18%: 237 ng m -3 in 1987-1988; 246 ng m -3 in 1988-1989; 264 ng m -3 in 1989-1990; and 223 ng m -3 in 1990-1991. The relatively low variability in nss SO42- may be a consequence of its multiple sources. The relatively high continental input during mid-December 1988 through mid-February 1989 (as reflected in the high NO? concentrations) is offset by a relatively low input from oceanic sources (as reflected in the low MSA concentrations). The relationships between nss SO ]-, NO?, and MSA are discussed in more detail later in this report.

4. Interstation Comparisons

In our previous reports, we compared the data from Mawson with data obtained by others at different high latitude locations, principally GvN (Wagenbach et al., 1988) and Cape Grim (Ayers et al., 1986). This comparison suggested that the concentrations and seasonal variability of nitrogen and sulfur species in aerosols may be relatively uniform over a broad sector of the Antarctic (Prospero et al., 1991; Savoie et al., 1992). To further investigate this uniformity, we compare the concentrations in samples collected during concurrent time periods at Mawson, Palmer, and Marsh. The data currently available for the latter two stations cover the time periods from 3 April 1990 to 15 June 1991 at Palmer and 28 March 1990 to 1 May 1991 at Marsh. Samples from Mawson and Palmer were available throughout their respective time periods, and hence the annual average concentrations of all of


the species at these stations can be directly compared (Table II). Because of the substantial local pollution sources at Marsh (as indicated in the Introduction), we discuss only the Na +, MSA, 21°pb, and 7Be data from that station. Because of an error during shipping, samples collected from 23 January to 3 April 1991 at Marsh never arrived for analysis at Miami although they were later analyzed at DOE for the radionuclides. As a consequence, our data do not contain a substantial portion of the summer portion of the annual cycle. Hence for the species that exhibit a significant seasonality, the simple average from this set could substantially misrepre- sent the true annual average. For more rigorous comparisons, we rely on the concurrent time series plots from all three stations (Figure 5) and examine the general seasonal trends in the concentrations at each.

4.1. Sea-Salt

The Na ÷ concentrations are indicative of the major differences in the direct influence of marine sources at the three sites. Mawson is obviously the least marine of the three sites. Of the 209 Na + concentrations measured at Mawson, all are less than 700 ng m -3 with all but 11 (5%) being less than 300 ng m -3. At Palmer, Na + ranges between 280 and 6700 ng m -3 but only 1 is less than 300 and only 15 (24%) are less than 700. The strongest effect of sea-salt aerosols is at Marsh where the 47 Na + concentrations range between 230 and 9100 ng m-3; of these, only 2 (4%) were less than 700 ng m -3. For the period from 3 April 1990 to 17 June 1991, the mean Na + concentration at Palmer (1200 ng m -3) is more than 8 times higher than that at Mawson (145 ng m -3) (Table II). The Na + concentrations at Marsh do not appear to exhibit a strong seasonal variation, and hence the simple average may be reasonably representative of the annual mean. If that is the case, then the mean Na + concentration at Marsh (3150 ng m -3) is more than 2.5 times higher than that at Palmer and more than 20 times higher than that at Mawson. The relative concentrations are comparable if we consider only those 47 weeks when samples were obtained concurrently at all three sites. For those periods, the mean Na + concen-

Table II. Average 12-month mean concentrations at Mawson and Palmer Station for the period 3 April 1990 to 17 June 1991, and at Marsh for the period 4 April 1990 to 3 April 1991

Station NO~ Na + NSS SOl- MSA NHI 21°pb (ng m -3 ) (ng m -3) (ng m -3 ) (ng m -3) (ng m -3) (fCi m -3 )

Mawson 30.4 (18.9) 145.0 (143.0) 89.6 (98.6) 23.2 (23.6) 16.0 (19.3) 0.87 (0.41) Palmer 18.1 (9.6) 1200.0 (830.0) 98.5 (121.0) 48.8 (89.2) 24.6 (41.8) 0.43 (0.24) Marsh - (-) 3150.0 (1920.0) - (-) 30.0 a (-) - (-) 0.46 (0.31)

Values in parentheses are the sample standard deviations. a This value is a very rough approximation; no MSA data are available for the middle to late summer when the concentrations are expected to be very high.

104 D. L SAVOIE ET AL.

t~ I E

< u3

t~ I E

(/)

CO Og z

I E o~

E "E o E E <

t,) I E

4 - - A Palmer i, I,I ~ - - ~ M a r s h a ~ I

t~ ]~,111

M a ' Ap ' M~-'-J 'G'-~J~ ~ ] b - ' Se ' Oc ' No ' pe ' Jo ' Fe ' lvia ' Ao ' Ma ' Ju

o--oNtawson '1 t a - - aPalmer I = t

Ma' / ,p ' M a ' J u ' Ju ' A u ' S o ' O c ' N o ' De' Ja ' r e ' M a ' A t m ' M a ' J u

200 180- 160 ~ 140. 120. 100 i 80- 60 40 i

20 0

400 350 300 250 200 150 1 O0

5 0 , 0

--50

120,

100'

80'

60.

40-

20.

0

90-

80~

70

60 i

50

40

30

20

I0~

0

%8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2.

0.0

o - - o M a w s o n - - a P a l m e r tI1¢

I l l t~ . / fll~

M a ' Ap ' Ma -~Ju ~ 'Ju- ' A u ' Se ' Oc ' No ' D~ ' Ja ' F'e ' M a ' Ap ' k ( a ' JG

O-- o Mawson a - - a P a l m e r

bla'Ap' Me'Ju ' Ju . . . . 'Au' SO'Oc'No' D~'Ja 'Fe Ma'Ap'bla'Ju

o - - o Mawson

A - - a P a l m e r /~

~ " o - - ~ M a r s h / I I ?E

0

~ ' A ~ ' Se'Oc ~ oN ' De' Ja ' Fe 'Ma Ap Ma Ju

Fig. 5. Time series plots of methanesulfonate (MSA), non-sea-salt (nss) SO~-, NH~, NO~, and 2mpb at Mawson, Palmer, and Marsh for the per iod March 1990 through June 1991.

NITROGEN AND SULFUR IN ANTARCTIC AEROSOLS 1 0 5

tration at Marsh (3220, s = 1870) was 3 times higher than that at Palmer (1080, s = 720), and 21 times higher than that at Mawson (151, s = 147).

The Na + concentrations at the peninsula stations are reasonably typical for coastal locations while those at Mawson are extremely low for such a site. The low Na + concentrations at Mawson are a consequence of the strong katabatic flow over eastern Antarctica which prevents a strong transport of sea-salt directly to Mawson by on-shore winds. Surface winds at Mawson are from inland Antarctica more than 80% of the time (Gras and Adriaansen, 1985; Ito, 1989; Streten, 1963). As a result, the four-year mean Na + concentration at Mawson is only about five times higher than the mean at South Pole (19 ng m-3; Tuncel et al., 1989).

In contrast to Mawson, there appears to be no significant seasonality in the sea- salt concentrations at the peninsula stations. At both Palmer and Marsh, strong Na + peaks occur throughout the year. However, the data from these stations are still rather limited, and certainly do not rule out the possibility that a significant seasonality would appear in a long term record.

4.2. Methanesulfonate, Non-Sea-Salt Sulfate, and Ammonium

MSA, nss SO 2-, and NH~ all share reasonably similar seasonal cycles at both Palmer and Mawson. Each of the species exhibits extremely low concentrations throughout the austral winter and then substantially higher concentrations during the austral summer. Despite these similarities, there are some notable differences.

Given the order of magnitude higher sea-salt concentrations at Palmer and Marsh in comparison to those at Mawson, one might expect the concentrations of MSA to be substantially higher as well. In fact, on an annual basis, the mean MSA concentration at Palmer (48.8 ng m -3) is more than a factor of two higher than that at Mawson. However, MSA exhibits a very strong seasonality at each of the three locations, and hence comparisons on shorter time scales may be more appropriate. Although the summer MSA data at Marsh are limited, the timing of the seasonality appears to be reasonably consistent at all three stations. At each of the stations, the concentrations exhibit decreasing trends during April and May 1990, begin to increase again in late September to high summer values, and then begin to decrease again in March 1991.

During the winter, the MSA concentrations at all three stations are very low and comparable to one another. For example, for the 22 samples that were collected at each station from May through September 1990, the means were 4.8 ng m -3 (s = 3.7) at Mawson, 4.1 (s = 4.1) at Palmer, and 4.4 (s = 5.5) at Marsh. During May and June 1991, the means were 6.5 ng m -3 (s= 2.8) at Mawson and 7.4 (s = 4.1) at Palmer. Although the 1991 means are somewhat higher, they do not differ significantly from those in 1990.

During the austral summer, the MSA concentrations are higher and far more variable, and large differences are evident between those at Mawson and the peninsula stations. The highest concentrations occur from December through March.

106 D.L. SAVOIE E T AL.

For the 17 samples collected at each station from December 1990 through March 1991, the mean MSA concentration at Palmer (122 ng m -3, s = 127) is 2.4 times higher than that at Mawson (51, s = 21). The highest concentration measured at Palmer (570 ng m -3) is three times higher than the highest concentration that we have measured during our nearly five years of sampling at Mawson, 170 ng m -3. For the more limited, seven-week period of December 1990 and January 1991 when samples were collected at all three sites, there is less disparity in the Mawson and Palmer concentrations. For that period, the mean concentration at Palmer (102 ng m -3, s = 5 6 ) is a factor of 1.6 higher than that at Mawson (64 ng m -3,

s = 25), and the mean at Marsh (84 ng m -3, s = 50) falls about midway between those extremes, i.e. a factor of 1.3 higher than at Mawson.

Overall, the seasonal cycle of nss SO ]- at Palmer is very similar to that at Mawson (Figure 5), and the concentrations at the two sites are comparable throughout their concurrent records. In fact, the mean nss SOl- concentration at Palmer (98.5 ng m -3) is only about 10% higher than the concurrent mean at Mawson (89.6, Table II), and this difference is not statistically significant. Moreover, the mean at Palmer is virtually identical to the 1988-1989 mean at Mawson (97.8, Table I).

The similarity between the nss SO ]- concentrations at Palmer and Mawson continues when the data are separated according to season. For example, for the period from May through September 1990, the mean nss SO42- concentrations are 5 ng m -3 (s = 34, n = 22) at Mawson and 23 (s = 38, n = 22) at Palmer. The apparent differences between the two stations are largely a consequence of two consecutive relatively high concentrations at Palmer and three very low concentrations at Mawson. If these are not considered, then the means become 15 ng m -3 (s = 22) at Mawson and 14 ng m -3 (S = 24) at Palmer. Nonparametric statistics also indicate that the nss SO]- medians and frequency distributions for this period at the two stations do not differ significantly at the 95% confidence level even when all of the samples are considered. Similarly, from April to mid-June 1991, the mean nss SO]- concentrations at Palmer and Mawson are nearly identical, 29 ng m -3

(s = 22, n = 10) and 30 (s = 25, n = 11), respectively. At both Mawson and Palmer, the nss SO42- concentrations begin to increase in early October to the much higher levels which occur during the summer and then begin to decrease again in March. As during the winter, the summer concentrations at the two stations do not differ significantly. For the three-month December through February period when the concentrations are the highest, the mean nss SO ]. concentrations 229 ng m -3 (s = 46, n = 13) at Mawson and 241 (s = 156, n = 13) at Palmer. The 5% difference in the means is clearly not significant although the variation at Palmer is considerably larger than at Mawson.

As with nss SO ]-, the timings of the seasonal cycles of NH] are very similar at Palmer and Mawson. At both sites, the NH] concentrations are consistently very low from April through September and exhibit strong peaks during the austral summer. However, while the NH] concentrations at Palmer are comparable to


those at Mawson during much of the year, they are considerably higher during the last half of the summer. In fact, it is the concentration difference during the latter period that accounts for the overall NH] mean concentration at Palmer, 25 ng m -3, being about 50% higher than that at Mawson for the same period, 16 ng m -3 (Table II).

During the low concentration period from the May through September 1990, the mean at Palmer, 3.1 ng m -3 (s = 2.8, n = 22), is about 25% lower than that at Mawson, 4.1 (s = 1.5, n = 22). This contrasts with the low concentration period from April through June 1991, during which the mean at Palmer, 5.5 ng m -3 (s = 2.2, n = 9), is a factor of 2 higher than that at Mawson, 2.7 (s = 1.3, n = 9). The reversal in order of the two sites and the relatively small differences during both periods suggests that the differences may be negligible for multi-year data sets. In fact, if the two periods are combined, the overall mean at Palmer, 3.8 ng m -3, is virtually identical to that at Mawson, 3.7 ng m -3.

The mean NH~ concentrations at Palmer and Mawson are also comparable during the beginning of the summer peak in November and December, 36 ng m -3 (s = 18, n = 9) and 40 (s = 21, n = 9), respectively. The noted similarities during most of the year contrasts sharply with the major differences during the later part of the summer from January through March. For this period in 1991, the mean NH~ concentration at Palmer, 64 ng m -3 (s = 66, n = 13) is more than 2.6 times higher than that at Mawson, 24 ng m -3 (s = 21, n = 13).

Notably, both nss SO 2- and NH~ exhibit very high concentrations at Palmer during the week of 19-25 January 1991, concurrent with a very high concentration of MSA. During this period, the concentrations of nss SO~- (660 ng m-3), NH~ (250 ng m-3), and MSA (580 ng m -3) are all factors of 3-5 higher than their November-January means. Moreover, the similarities and differences between the seasonal cycles and concentration levels of NH~ at Palmer and Mawson are closely matched to those of MSA. For both species, the concentration levels from January through March are factors of 2-3 higher at Palmer than at Mawson; in contrast to November and December when the levels at the two stations are more nearly equal. The similarities between NH~ and MSA might indicate that the primary source for the NH] or, more likely, its precursors, is the same as for MSA, i.e. biological activity in the ocean.

4.3. Nitrate and Lead-210

In comparisons between Mawson and Palmer, NO~ and 21°pb contrast sharply with the other constituents that we have measured. For both of these species, the annual average concentrations at Palmer are a factor of about two lower than those at Mawson. The mean NO~ concentration at Palmer, 18.0 ng m -3, is a factor of 1.7 lower than that at Mawson, 30.4 ng m -3. The absence of a substantial seasonal cycle in 21°pB at Palmer (Figure 5) suggests that the average for Marsh may be reasonably representative of the annual average despite the lack of data from the


late summer. If this is the case, then the m e a n 21°pb concentrations at Palmer and Marsh are virtually identical, 0.44 and 0.46 fCi m -3, respectively, and nearly a factor of two lower than that at Mawson, 0.86 fCi m -3. In fact, the means at the peninsula stations are 20% lower than the lowest composited monthly mean at Mawson, 0.55 fCi m -3 during June.

As with many of the other species discussed previously, the mean concentrations of NO~ at Palmer and Mawson are very similar during the periods when the concentrations are at a seasonal minimum. This is true for the period from April to mid-June 1990 and also that from March to mid-June 1991. The respective means at Palmer and Mawson are 10.7 ng m -3 (s = 8.9, n = 12) and 10.4 (s = 5.0, n = 12) during the former period and 14.5 ng m -3 (s = 6.2, n = 15) and 14.4 (s = 5.3, n = 15) during the latter. However, during the remainder of the record, the concentrations at the two stations differ substantially. The difference is most apparent during the two-month mid-October to mid-December NO 3 peak at Mawson when the NO~ mean at Mawson, 63 ng m -3 (s = 13, n = 7), was about three times higher than that at Palmer, 22 ng m -3 (s = 8, n = 7). However, even for the periods from July through mid-October and mid-December to early February, the means at Mawson, 32.3 ng m -3 (S = 9.9, n = 16) and 39 ng m -3 (S ~ 14, n = 8), are about twice the respective means at Palmer, 16.8 ng m -3 (s = 6.5, n = 16) and 22 ng m -3

(s = 13, n = 8). Despite the substantial differences in the absolute concentrations, the timing of

the NO~ seasonal cycles at the two stations are reasonably similar. At both sites, the seasonal minimum occurs during May and June and the concentrations then begin to increase during July. However, at least during 1990, the concentrations at Palmer do not exhibit the rather strong peak from mid-October through early December that occurs in every year that we have sampled at Mawson.

Substantial uncertainty occurs when attempting to compare the 21°Pb concentrations at Palmer and Marsh with those at Mawson for time periods of less than a year. Because the 21°pb concentrations at Marsh and Palmer are very low and near the detection fimit, the results from numerous samples must be averaged together to reduce the uncertainty to a level that would make such comparisons statistically meaningful. For the same reason, it is not possible with the current data set to dis- cern any significant seasonal cycle in 2~°pb at these locations. In this regard, we note that even at Mawson, where the mean concentrations are a factor of two higher, the seasonality could not be reasonably deduced from the data for a single

year.

4.4. Summary of the Interstation Comparisons

Our results indicate that there are some notable similarities and differences between the Antarctic peninsula stations at Palmer and Marsh and the east Ant- arctica station at Mawson. As one might expect based on the differences in the local meteorology, the sodium concentrations at Palmer and Marsh are about an


order of magnitude higher than those at Mawson and more typical of those expected at coastal locations. The concentrations of nss SO ]-, MSA and NH~ at Palmer are comparable to those at Mawson during the austral winter and into the early part of the austral summer. In fact, the nss SO 2- concentrations at Palmer are comparable to those at Mawson throughout the year. In contrast, there is an increasing disparity between the MSA and NH~ concentrations at Palmer and Mawson as the summer progresses with the concentrations during the late summer being factors of about 2.5-3 higher at Palmer than at Mawson. The differences dm:ing the late summer are, in fact, the primary reasons for the differences in the annual means at the two stations. In contrast, the mean concentrations of NO? at Palmer and of 21°pb at both Palmer and Marsh are factors of two lower than those at Mawson. Moreover, the NO3 concentrations at Palmer do not exhibit an austral spring maximum which is a persistent feature at Mawson. However, as with nss SO]-, MSA, and NH], the winter minimum concentrations at the two stations are about equal and occur during the same time of year, about mid-April to mid-June.

5. Non-Sea-Salt Sulfate, Methanesulfonate, Nitrate Relationships

5.1. Mawson

In our previous study of the results from Mawson (Savoie et al., 1992), we showed that seasonally higher concentrations of NO3 were associated with lower MSA/nss SO~- ratios. This feature continues to persist through the longer data record (Figure 6). These results indicate that the depressions in the MSA/nss SO~- ratio might be a consequence of the input of nss SO~- from an additional source and that the magnitude of this input might be directly related to the concentration of NO~. Multiple variable regression of the nss SO42- data using NO~ and MSA as the independent variables yields the following equation:

Nss SO 2- = 1.59(_+ 0.17) x NO~ + 2.80(_+ 0.13) x M S A - 12.1 (1)

where all of the concentrations are in ng m -3, N = 244, r 2 = 0.73, and the standard error in the estimated nss SO ]- concentration (SEnss) is 49 ng m -3. If the negative intercept is attributed to 'background' NO3 which has no nss SO ]- associated with it, then the equation is appropriately rewritten as:

Nss SO42- = 1.59 x (NO 3 - 7.6) + 2.80 x MSA (2)

A value of 7.6 ng m -3 for the NO 3 'background' is reasonably consistent with the lowest composited monthly mean NO~ concentration; the mean for all NO3 concentrations during May is 9.8 ng m -3.

The regression coefficient for MSA indicates that the nss SO2-/MSA ratio re- suiting from the oxidation of DMS is 2.80 _+ 0.13. However, the value of this coefficient is primarily a function of the concentrations during the summer when the concentrations of both MSA and nss SO ]. exhibit their largest absolute variance.

110 D . L . SAVOIE E T AL.

69 C" u~ 03

¢- ° _

r " ¢-

rY

© (9

I cO

fib I E

c-- v

03

~7 q - - m

09

09 LO : 7

4 0 0

350 300 250 200 150i 10o-

5 0 i

ol

S g "! ~ 1

; , ! tt',fI/ I N }/

" i b ~ ) ~ i b 8 8 I 4 ~ 8 9 ' ' I . . . . b ~ b ' ' I ' ' i b g i ' '

o MSA/NSS S04=, 1000;

4 0 0

I i

m

~ NSS S04-; " , 'N03- .2

350 300 250 . ' ~ ~ 200 -: ~ '~ 150' ~ o o ~o 100 o o

o o (

50 o o

0 . . . . . . . . . . . i . . . . . . . . . . . i . . . . . . . . . . . i . . . . . - . . . . . . i . . . . . . . . . . . i

o

q

o o ° ~

o c

) o o

o

1 9 8 7 I 1988 1 1989 I 1990 I 1991 1 - - Measured o Regress ion Est imate

Fig. 6. Plots of (a) the 8-week running means of the M S A / n s s SO ~- mass ratio at Mawson and of the nss SO42- and NO? concentrations and (b) a c o m p a r i s o n of the measured weekly nss SO~- concentrations and those calculated from the regression equation with MSA and NO~- as the independent variables (see text for a discussion of the equation).

To check the seasonal constancy of this ratio, we separately consider only those samples for which the NO 3 concentration is less than 15 ng m -3. For these 72 samples, the mean MSA and nss SO]- concentrations, 10.5 ng m -3 ( s = 11.3) and 30.9 ng m -3 ( s = 34.3), yield a nss SO]-/MSA ratio of 2.93 which is within 5% of that obtained from the regression analysis. The close agreement between the ratios indicates both the appropriateness of the regression coefficient and the seasonal invariance in the MSA/nss SO ]- ratio resulting from the oxidation of DMS.

The regression coefficients obtained using the four-year data set differ some-

N I T R O G E N A N D S U L F U R IN A N T A R C T I C A E R O S O L S 111

what from those which we obtained earlier using only 2.5 years of data. For example, the MSA coefficient, 2.80, from the longer set is 13% lower than the previous value of 3.22. The nitrate coefficient decreased by 6%, from 1.7 to the current 1.6. However, the nss SO 2- concentrations predicted from MSA and NO3 using the current regression equation agrees reasonably well with the measured nss SO 2- concentrations over the entire Mawson data set (Figure 6). The large number of data used in this analysis imparts substantial inertia to the current estimates of the regression coefficients, and, hence, it is unlikely that they will change significantly in the near future.

On the basis of these coefficients, we estimate that, for the four-year period, about 71% (64 ng m -3) of the total mean nss SO 2- (90 ng m -3) is associated with MSA. The remaining 29% (26 ng m -3) is associated with non-background NO~. These percentages are virtually identical to those that we reported for the shorter term data set, 72% associated with MSA and 28% with NO3. As we have previously shown, the seasonality of the NO3 concentrations is most similar to that of 21°pb. Moreover, as we shall show later, the spatial distribution of NO 3 is also most similar to that of 21°pb. These results support our previously published conclusion that the NO3 is probably derived predominantly from a continental source (Savoie et al., 1992).

5.2. Palmer Station

The data from Palmer provide for a preliminary assessment of the spatial variability of the nss SO2--MSA-NO3 relationship in the Antarctic region. Toward this end, we consider (1) the comparability of the regression parameters using the data from Palmer versus those from Mawson and (2) the accuracy with which the equation for Mawson can predict the nss SO 2- concentrations at Palmer.

Multiple variable regression of the Palmer nss SO 2- data using MSA and NO3 as the independent variables yields the following equation:

Nss SO 2- = 0.65(_+ 0.47) x NO 3 + 1.70(+ 0.10) x MSA+ 8.5 (3)

where all concentrations are in ng m -3, N = 61, r 2 = 0 . 8 6 , and SEns s = 33 ng m -3. However, in contrast to that for Mawson, the intercept is insignificant. Regression with a forced zero intercept yields:

Nss SO~- = 1.02(+ 0.29) x NO 3 + 1.71 (_+ 0.10) x MSA (4)

w i t h r 2 = 0.86 and SEns s = 33 Na +, i.e. identical to those of Equation (3). In fact, the nss SO 2- concentrations predicted by these two equations are virtually indistinguishable (Figure 7).

In Figure 7, we also show the nss SO 2- concentrations predicted on the basis of the equation derived for Mawson. During the winter months when the concentrations of all of the species are relatively low, the estimates from all three equations are reasonably consistent and the vast majority of the estimates are within the


I000 I

" - " 9 0 0 " r ~

I 8oo F: 700

c- 600

o 500 c~ 400

-5 500 Of) 200 Lf) co 100- Z

0

-100

A NSS S 0 4 = = 0 . 6 5 . N 0 3 + 1.70*MSA + 8.5

v N S S S 0 4 = = 1.02.N03 + 1.71*MSA

o NSS S04 = = 1 .59 .N03 - + 2.80*MSA - 12.1

o

0 ~ 1 1 0

0 O0 0 0

- - -~v v

o

o

i i i i i i i 1 [ i i i l ,

Ap Ma Ju Ju Au Se Oc No De da Fe Ma Ap Ma Ju

Fig. 7. Plot of the measured nss SOl- concentrations (solid line) at Palmer Station and the estimated concentrations calculated from regression equations based on the Palmer data (open triangles) and from the regression equation based on the Mawson data (open circles).

uncertainties of the measured nss SO 2- concentrations. However, major differences between the Mawson and Palmer equations are clearly evident in the spring and summer when the MSA and nss SO ]- concentrations are much higher. During this period, the estimates based on the Mawson equation are usually much too high. They are closest when the MSA concentration is at a relative minimum. This suggests that the MSA coefficient from the Mawson regression is too high for Palmer. The MSA coefficient calculated for Palmer, 1.71, indicates that the nss SO]-/MSA ratio from DMS oxidation for that station is about 40% lower than that at Mawson, 2.80. Notably, the NO? coefficient for Palmer in Equation (4) is also about 40% less than that at Mawson. Because the sample set at Palmer is still relatively small, the standard error in the regression coefficient may not truly reflect the real uncertainty. However, the differences are large enough that they are unlikely to have occurred by chance alone.

Based on the coefficients in Equation (4) and the average concentrations in Table II, the marine biogenic source is estimated to account for about 85% of the total mean nss SO ]. concentration at Palmer. This percentage is somewhat larger than that suggested previously for Mawson, 76%. Hence, while Mawson is geo- graphically more remote from continental sources than Palmer, a significantly larger percentage of its nss SO ]- appears to be derived from non-marine sources.

6. Ammonium-Acidic Species Interrelationships

The acidity of the aerosols can have a strong impact on the solubility of many aerosol constituents, e.g. trace metals such as iron. Consequently, it is instructive to con-

NITROGEN ANT_) SULFUR IN ANTARCTIC AEROSOLS 113

sider the degree to which the acid species in the Antarctic region are neutrafized by NH3 which, apart from the CO32-/HCO? associated with sea-salt, is usually the major base.

If one assumes that all of the NH] in the aerosols results from the uptake of gaseous NH3, then the four-year means indicate that, on average, ammonia neutralizes about 40% of the total acid derived from H 2 S O 4 , MSA, and H N O 3 at Mawson. At Palmer, the annual neutralization is about 20% higher, about 48% of the total acid. However, these values may be somewhat misleading. Particle size distribution studies in the marine environment have shown that particulate NO 3 exists primarily in the sea-salt fraction, indicating that H N O 3 is neutralized by reaction with sea-salt carbonate and by reaction with NaC1 and subsequent release of HC1. These studies further indicate that NH~ rarely exists above detectable levels in these supramicron particles (Berresheim et al., 1990, 1991; Zhu et al., 1992). Consequently, only H 2 S O 4 and MSA are likely to be neutrafized by reaction with N H 3 .

On average, reactions with N H 3 neutralizes about half of the acidity derived from H2SO 4 and MSA in our samples: 49% at Mawson and 53% at Palmer. How- ever, because the seasonal cycles of NH], nss SO24 -, and MSA are not identical, the NH~/(nss SOl- + MSA) equivalent ratio displays a significant annual cycle. As one might expect, the greatest monthly average neutralization (56%) at Mawson occurs during December, concurrent with the peak in NH~. For December 1990, the percentage at Palmer is similar, about 58%. During January at Mawson, NH~ usually decreases more rapidly than nss SOl-, and MSA increases to its seasonally highest level. Consequently, the NH~/(nss SO42- + MSA) ratio typically decreases at this tame. The lowest degree of neutralization at Mawson (about 30%) occurs during February and March and is a factor of about two lower than the maximum. In contrast, at Palmer, the neutralization during January and February (61% and 62%, respectively) is virtually the same as in December.

The apparent neutralization in samples collected aboard aircraft over the southern ocean during December 1986 is somewhat higher than those at Palmer and Mawson. Berresheim et al. (1990) reported a mean NH~/nss SO 2- molar ratio of 1.87 in the marine boundary layer which yields an equivalent ratio of 0.94. With a mean MSA/nss SO ]- molar ratio of 0.4-0.6, the NH~/(nss SO42- + MSA) equivalent ratio is about 72-78%. Given the short time spans over which these aircraft measurements were made, these results are certainly not inconsistent with our data and clearly support our conclusion that a large percentage of the aerosol acidity is neutralized by reaction with N H 3.

7. Discuss ion

The results from this study suggest two major points which require more detailed discussion: (1) the spatial distributions of MSA and the nss SO42-/MSA ratio; and (2) the spatial distributions of NO 3 and 21°pb.


Z1. Spatial Distribution of MSA and Biogenic Nss SO~ /MSA Ratios

On an annual basis, the mean MSA concentrations at Palmer are a factor of about two higher than those at Mawson, suggesting that the means may be higher at stations that are directly affected by transport from the surrounding oceans. However, currently available long-term data are not consistent in this regard. For example, the annual mean MSA concentration at Georg yon Neumayer (GvN), 39 ng m -3

(Minikin and Wagenbach, 1990), is about 70% higher than that at Mawson and 20% lower than that at Palmer. Although GvN is also a coastal site in east Ant- arctica, the katabatic flow there is considerably weaker than at Mawson and the winds at GvN are frequently from the ocean. Consequently, the mean Na + concentration at GvN is 4 times higher than at Mawson, but still more than 3 times lower than at Palmer. On the other hand, the mean MSA concentration at Cape Grim is nearly identical to that at Mawson (Ayers et aL, 1986, 1991) even though Cape Grim consistently experiences very strong flow from the ocean as evidenced by the location's high Na ÷ concentrations. Clearly, the evidence from the available MSA data is still ambiguous, and the spatial variability in the mean MSA concentration may simply reflect the spatial heterogeneity in the emission flux of DMS, in the relative concentrations of its oxidation products and/or in their atmospheric transport and removal.

There appears to be a significant difference between the nss SO]-/MSA ratio derived from DMS oxidation at Mawson and that at Palmer. The MSA regression coefficients presented previously indicate that this ratio is about 1.71 at Palmer compared to about 2.80 at Mawson. These results suggest that the ratio in aerosols may be substantially lower over the southern ocean than that in surface air from the interior of Antarctica. As stated earlier, the air sampled at Mawson is predominantly from the interior of the continent because of the strong and persistent katabatic flow.

Although data from other studies may support this conclusion, direct comparisons of our results with those of others is difficult because, in most other studies in the Antarctic, nss SO4 2- has simply been assumed to be derived exclusively from the oxidation of DMS. Consequently, the ratios from the other studies must be considered as the maximum to be expected from DMS oxidation alone. Moreover, the uncertainties in the nss SO ]- concentrations were very high in many of the previous studies as a consequence of high sea-salt SO42- concentrations. Despite these potential problems, the nss SO2-/MSA ratios in shipboard samples in the Drake Passage and Gerlache Straits were comparable to those at Palmer, 1.30-1.89 (Pszenny et aL, 1989; Berresheim, 1987). Ratios in aircraft samples in the marine boundary layer over the southern ocean were somewhat higher, 1.9-3.1 (Berres- heim et al., 1990). Based on the data of Wagenbach et al. (1988) and Minikin and Wagenbach (1990), the ratio at GvN is similar to that at Palmer and over the southern ocean, about 1.8.

The nss SO]- /MSA ratios which have been measured in higher altitude inland


Antarctic ice cores are also much higher than those from coastal snow/ice cores. The lowest reported mean ratio is 2.17 in a core from the low altitude (398 m asl) site on Dolleman Island (Mulvaney et al., 1992). A somewhat larger mean ratio (2.56) was measured in a core from Law Dome (Ivey et al., 1986). At higher altitude sites on the Antarctic peninsula the mean ratios increase to 2.70 at Gomez Nunatak (1130 m asl) and 3.12 on the Dyer Plateau (1900 m asl; Mulvaney et al., 1992). If a majority of the nss SO42- in these cores is indeed derived from DMS oxidation, then these means are reasonably consistent with those in aerosols over coastal Antarctica and the southern oceans. By comparison, the mean nss SOl-/ MSA ratios are sharply higher in cores from the interior of Antarctica: about 20- 25 in the most recent sections from Vostok, Dome C (Legrand et al., 1991), and the Dominion Range (Whung, 1991).

There are several potential reasons for the spatial variations in nss SO]-/MSA ratios that have been observed over and around Antarctica. The relative amounts of MSA and SO 2 produced from the oxidation of DMS could vary substantially over the region. However, the seasonal consistency of the ratio at Mawson and reasonably consistent ratios measured aboard ship, aboard aircraft in the MBL, and at stations directly impacted by air flow from the ocean indicate that this is probably not a viable explanation.

The nss SO]-/MSA ratios at Mawson and Palmer and over the southern ocean, in general, are factors of 5-10 lower than those of about 15-20 found over the tropical and subtropical oceans (Saltzman et al., 1985; Savoie et al., 1993a, b). The lower ratios at high latitude sites have frequently been attributed to the increased importance of the addition of OH relative to H abstraction in the oxidation of DMS as the temperature decreases (Hynes et al., 1986). However, the oxidation of DMS to SO2 and MSA involves a complex sequence of reactions which are still poorly understood; the attack of DMS by OH is only the initial part of this sequence. If the primary cause of the ratio variations were simple temperature dependence, the ratio at Mawson and Palmer should be highest during the mid-summer (December through February) when the temperatures are highest. Moreover, the ratio over the Antarctic continent should be lower than that in the MBL of the southern ocean and not higher as our results and those of others clearly indicate.

Higher ratios over the interior of Antarctica could result from a larger relative amount of nss SO ]- from sources other than DMS oxidation. The results from our regression analyses indicate that this is unlikely to be the cause of the differences in the ratios between Mawson and Palmer. However, it could be a major cause of the sharply higher ratios in the interior ice cores. Because of the very low MSA concentrations in those cores, even small additions of nss SO]- from other sources could have a substantial impact on the overall ratio. Our results from Mawson and Palmer indicate that significant quantities of nonbiogenic nss SO4 > occur at both locations. The correlation of this nonbiogenic fraction with NO 3 suggests that this portion may be derived from continental sources.

Another potential explanation is that the differences in the ratios are a conse-


quence of a more rapid loss of MSA compared to nss SO42- during transport of air parcels into the interior of Antarctica. Such a phenomenon has some rather inter- esting implications and, in fact, could contribute to the spatial variations both in the nss SO2-/MSA ratio and in the MSA concentration in Antarctic snow and ice. The results from the Vostok and Dome C cores (Legrand et al., 1991; De Angelis et al., 1987; Petit et al., 1981) show that the concentrations of MSA, nss SO~-, sea-salt, and continental dust all increase substantially during the cold glacial periods. These authors generally conclude that the higher concentrations reflect a stronger atmospheric circulation and a faster aerosol transport over the Antarctic continent. Inter- estingly, the nss SO2-/MSA ratio also decreases substantially during these periods, from values of about 20 during the interglacials to about 7. Even at its minimum, however, the ratio remains considerably greater than that at Mawson and certainly much greater than that over the open ocean. Conceivably, the more rapid transport of material into the interior of Antarctica could reduce the loss of MSA relative to that of nss SO 2-. This scenario would result in lower nss SO42-/MSA ratios over the interior without the necessity of invoking major changes either in the actual ratio produced by DMS oxidation or in the percentage of nss SO 2- from other sources.

A potential mechanism for a more rapid loss of MSA relative to nss SO 2- involves differences in their physicochemical properties. Few studies have been made of the particle size distributions of MSA and nss SO 2- in and around Antarctica. However, at Cape Grim, the nss SO2-/MSA ratios in bulk aerosols (Ayers et al., 1986) are about a factor of two lower than those in the submicron fraction alone (Ayers et al., 1991). These results strongly suggest that the mass median diameter of MSA in this region is significantly larger than that of nss SO42-. Similar conclusions have been drawn with regard to their relative particle size distributions in tropical and subtropical regions (e.g., Pszenny et al., 1992; Saltzman et al., 1983; Zhu et al., 1989). The larger MSA containing particles could be more rapidly removed from the atmosphere by both dry and wet processes. However, available size-distribution data are still too sparse and our knowledge of the dry deposition rates as a function of particle size and atmospheric conditions is still too crude to make any definitive assessment of the magnitude of this potential effect.

While the latter explanation may, in some ways, be more esthetically pleasing than others, the real reason for the variation in the nss SO]-/MSA ratio needs to be determined if we are to accurately interpret the ice core records. There are clearly numerous other possible explanations, including differences between the oxidation pathways of DMS in the free troposphere and those in the underlying marine boundary layer. Recall that the particles over Antarctica arrive there via the free troposphere and some, if not all, may actually be formed in that part of the atmosphere. Notable in this regard is the fact that we still do not know, for certain, the reason that the nss SO4Z-/MSA ratios in the high latitude marine boundary layer are substantially lower than those in tropical regions. Clearly, long term measurements of MSA and nss SO ]- over the interior of Antarctica would be especially helpful in resolving this issue, particularly if these were combined with ancillary measure-


ments which could be used to assess the potential impact of nss SO 2- sources other than DMS oxidation. A definitive resolution of this problem is crucial to the correct interpretation of ice core records. Furthermore, the data that would result from such a study would be extremely useful for validating the results of general circulation models of the atmospheric sulfur cycle such as that being used by Langner and Rodhe (1991).

7.2. Spatial Distributions o f N O 3 and 21°pb

That the NO~ concentrations at Palmer during the late spring and summer are about a factor of two lower than those at Mawson is intriguing in view of our previous conclusion that the dominant source of the NO~ at Mawson is probably continental. Since Palmer is much closer to a potentially major continental source, South America, one might have reasonably expected the opposite. It is pertinent in this regard that the NOB concentrations at Macquarie Island (54 ° S, 159 ° E) are also lower than those at Mawson and GvN (G. Ayers, CSIRO, Australia, personal communication). These results indicate that there might be a relative minimum in the marine boundary layer (MBL) NO3 concentrations between Antarctica and the mid-latitudes.

The lower NO~ concentrations at Palmer and Macquarie are consistent with our contention that the NOB concentrations in this region are most strongly related to those of 21°pb. Combined results from a number of sources indicate that 21°pb (as well as 222Rn) in the MBL exhibits a relative minimum between 40 ° and 60 ° S (Polian et al., 1986; Lambert et al., 1990). Our results clearly support this conclusion. For example, the mean concentrations at Palmer and Marsh are 0.43 and 0.46 fCi m -3, respectively. Figure 3 of the report by Lambert et al. shows that the average concentrations of 21°pb in the region of the minimum are about 0.4 fCi m -3 or less. Concentrations shown by Lambert et al. for east Antarctica are about 2 to 3 times higher and are consistent with our mean of 0.81 fCi m -3 for 21°pb at Mawson and with an essentially identical mean at GvN, 0.84 fCi m -3 (Wagenbach et al., 1988).

The reason for what appears to be a relative minimum in both NO~ and 21°pb between Antarctica and the mid-latitude continental areas remains uncertain. However, Polian et al. (1986) and Lambert et al. (1990) have proposed a reasonable explanation for the 2~°pb distribution which may also apply to NO B. The authors conclude that the transport of materials from the mid-latitude continental areas to Antarctica occurs primarily in the free troposphere. Such a scenario is highly probable in view of the fact that most long-range atmospheric transport occurs in the free troposphere (e.g., Knap, 1990) where the residence times are substantially longer than in the near surface layers. Recent results from a general circulation model indicates that the concentration of continental dust is also substantially higher in the free troposphere over Antarctica than in the underlying surface layer (Genthon, 1992). Subsidence of air over Antarctica resulting from radia-


tional cooling and subsequent drainage flow at the surface could effectively transport this material to lower altitudes. Hence, the transfer between the free troposphere and the surface layer may be much more effective over Antarctica than over much of the southern ocean. In addition, the total (wet plus dry) deposition rate in the marine boundary layer is likely to be significantly higher than that over the continent.

Legrand and Kirchner (1988), among others, have speculated that large nitrate particles associated with polar stratospheric clouds (PSC) could sediment into the troposphere and, hence, serve as a major source for nitrate in Antarctic ice cores and in the lower Antarctic troposphere. This source together with the transport of nitrate south from continental areas could be invoked to explain the nitrate minimum at 40°-60 ° S. However, as we noted in our previous report (Savoie et al., 1992), Iwasaka and Hayashi (1990) report that stratospheric aerosol enhancement at Syowa, occurs from June to September. Sighting frequencies of PSC's between 16 and 20 km altitude in the Antarctic increase from about 10% in June to a maximum of 60% in August and then decrease to 10% by early October (Poole et al.,

1991). In contrast, June through September is the period of the minimum NO 3 concentrations at Mawson, GvN, and Palmer. The seasonal cycle of NO3 at Maw- son also differs dramatically from that of 7Be which exhibits a peak in March. Our conclusion that the primary source of the nitrate is continental tends to support the later conclusion of Legrand and Kirchner (1990) that the major nitrate sources are located in the troposphere.

The source of the 'background' NO3 at Mawson remains very uncertain. Poten- tial sources are downmixing from the stratosphere, lightning, and the oxidation of ammonia or of amino acids associated with proteinaceous material. Savoie et al.

(1989b) showed that the mean concentrations of NO3 were very uniform at a level of about 100 ng m -3 o v e r the tropical South Pacific. At American Samoa, the NO3 concentrations were most strongly related to 21°pb suggesting that the principle source might be continental. However, the real source has yet to be rigorously investigated. The results from general circulation model calculations (e.g., Levy and Moxim, 1989; Penner et al., 1991) have provided little additional information in this regard since they simply do not indicate a 'background' concentration of the magnitude that was measured. Clearly, however, there must be either a source that has not been considered, a significant underestimation in the magnitude of one or more of the sources that has been considered, or a significant overestimation in the rate that the NO? is removed from the atmosphere in the models. In short, the principle source of the 'background' nitrate is an issue that needs to be investigated in far more detail than it has in the past.

8. Conclusions

In our earlier comparisons of our data from Mawson with those of others from a variety of locations in the Antarctic, we concluded that the aerosol composition


may be relatively uniform over a broad sector of the Antarctic. With our additional data from Mawson and now from Palmer and Marsh, that conclusion appears to remain valid in many respects but certainly not in all.

For most of the species that we have investigated, the timing of the seasonal cycles appear to be very consistent throughout the region of the southern ocean. This is particularly true for nss SO~-, MSA, and NH~. At all of the stations, the concentrations of these species exhibit large seasonal variations with concentrations during the austral summer that are more than an order of magnitude higher than those during the winter. For nss SO42- and NH~, even the absolute magnitudes of the concentrations are reasonably constant. The latter feature is especially inter- esting when one considers (1) that most of the nss SO~- is believed to be derived from the oxidation of DMS emitted from the ocean and (2) that the similarity for nss SO~- extends to the Antarctic sites that are subject to katabatic flow from the interior of the continent as well as to those that are impacted by direct boundary layer flow from the ocean. The concentrations of MSA appear to be somewhat more variable than the more limited data had previously suggested. Even so, the variations over the oceans and the coastal Antarctic stations are fairly well con- strained, being essentially invariant during the winter minimum and probably within a factor of two during the summer maximum. Considering that the annual mean MSA concentration at Mawson itself varies from 19 to 28 ng m -3, a regional variation of less than a factor of two is certainly within reasonable expectations.

Recently, Langner and Rodhe (1991) compared the results from their global three-dimensioinal model to the nss SO42- data that we had previously obtained at Mawson. The annual average concentrations from the model agreed very well with that calculated from the measured concentrations. However, the seasonal cycle in the model results was considerably weaker than the measurements indicated. The authors suggested that one possible reason for the discrepancy was that the strong katabatic flows from the interior of Antarctica at Mawson were not resolved by the model and might play an important role. However, our results indicate that the strength of the seasonality at the peninsula stations where katabatic flow is not a major feature is comparable to, if not stronger than, that at Mawson. Hence it is more likely that the discrepancy is a consequence of the alternative reason proposed by Langner and Rodhe, that the seasonal cycle of the DMS emissions from the southern ocean is much too weak in their model.

Spatial variations in the nss SO42-/MSA ratios are significant both in aerosols and in ice cores. The two most likely causes of these variations are (1) variations in the percentage of nss SO~- that is derived from sources other than DMS oxidation, and (2) more rapid removal of MSA relative to nss SO42- during atmospheric transport. Discerning which of these (if either) is the real cause is of major importance for understanding the tropospheric chemistry and atmospheric transport over a relatively large area of the earth. A correct evaluation of this issue is critical for ensuring that the transport and chemistry simulated in large scale models is correct or for assessing what needs to be done to correct them. It is also important in the


interpretation of ice core chemistry; the interpretations differ markedly depending on which of these phenomena is considered more significant. In this regard, long- term aerosol chemistry measurements over the Antarctic interior could be extremely valuable.

The primary source for NH~ in marine aerosols continues to be a subject of con- siderable research activity. NH~ appears to be a major component of aerosols over the southern ocean and its presence is important to the physical and chemical properties of the submicron aerosol particles. The consistent concurrence of the seasonal cycles of NH~ and those of MSA and nss SO 2- strongly suggests that the NH~ source is related to marine biological activity. What nitrogen species and atmospheric processes are involved in the production of the NH~ are not at all clear and certainly need to be investigated.

Our results continue to indicate that NO~ is most strongly related to 2~°pb and, hence, to a continental source. The low concentrations of NO 3 at Palmer compared to those at Mawson were unexpected but are consistent with the lower concentrations of 21°Pb. Our results provide additional evidence of a 21°pb minimum in the region of 40°-60 ° S (Polian et al., 1986; Lambert et al., 1990). However, the ramifications of this minimum with regard to atmospheric transport in the region need to be considered in more detail. So also does the origin of the 'background' NO? at Mawson.

Acknowledgements

We thank the Australia National Antarctic Expeditions (ANARE) for establishing, maintaining and operating our program at Mawson and J. Gras and G. Ayers, Commonwealth Scientific and Industrial Research Organization (CSIRO), Epping, New South Wales, Australia, for their assistance in this effort. We also gratefully acknowledge the invaluable assistance of M. Butler at Palmer Station and of A. Zuniga and M. Vazquez (Centro Meterologica Antarctico) and col. G. Palacios (Director, Direccion Meterologica de Chile) at Marsh, King George Island. The aerosol samples at all of the stations were collected and analyzed under U.S. Department of Energy (DOE) contracts DEAC1788EV90106 and DEAC1791EV90116.

References

Ayers, G. R, Ivey, J. R, and Goodman, H. S., 1986, Sulfate and methanesulfonate in the maritime aerosol at Cape Grim, Tasmania, J. Atmos. Chem. 4,173-185.

Ayers, G. R, Ivey, J. R, and Gillett, R. W., 1991, Coherence between seasonal cycles of dimethyl sul- phide, methanesulphonate and sulphate in marine air, Nature 349,404-406.

Berresheim, H., 1987, Biogenic sulfur emissions from the subantarctic and Antarctic oceans, J. Geo- phys. Res. 92, 13245-13262.

Berresheim, H., Andreae, M. O., Ayers, G. R, Gillett, R. W., Merrill, J. T., Davis, V. J., and Chameides, W. L., 1990, Airborne measurements of dimethylsulfide, sulfur dioxide, and aerosol ions over the southern ocean south of Australia, J. Atmos. Chem. 10, 341-370.


Berresheim, H., Andreae, M. O., Iverson, R. L , and Li, S. M., Seasonal variations of dimethytsulfide emissions and atmospheric sulfur and nitrogen species over the western North Atlantic Ocean, Tellus 43B, 353-372.

De Angelis, M., Barkov, N. I., and Petrov, V. N., 1987, Aerosol concentrations over the last climate cycle (160 kyr) from an Antarctic ice core, Nature 325,318-321.

Genthon, C., 1992, Simulations of desert dust and sea-salt aerosols in Antarctica with a general circulation model of the atmosphere, Tetlus 44B, 371-389.

Gibson, J. A. E., Garrick, R. C., Burton, H. R., and McTaggalt, A. tL, 1988, Dimethylsulfide concentrations in the ocean close to the Antarctic continent, Geomicrobiol. Z 6, 179-184.

Gibson, J. A. E., Garrick, R. C., Burton, H. R., and McTaggart, A. R., 1990, Dimethylsulfide and the alga Phaeocystispouchetti in antarctic coastal waters, Mar. Biol. 104,339-346.

Gras, J. L., and Adriaansen, A., 1985, Concentration and size variation of condensation nuclei at Mawson, Antarctica, Jr. Atmos. Chem. 3, 93-106.

Ito, T., 1989, Antarctic submicron aerosols and long-range transport of pollutants, Ambio 18, 34-41. Ivey, J. R, Davies, D. M., Morgan, V., and Ayers, G. E, 1986, Methanesulphonate in Antarctic ice,

Tellus 38B, 375-379. Iwaska, Y., and Hayashi, M., 1990, Effect of polar stratospheric clouds on geochemical cycle of

stratospheric material - stratospheric particle precipitation, Proc. NIPR Syrup. Polar Meteorol. glaciol. 3, 23-35.

Knap, A. H., 1990, The Long-Range Atmospheric Transport of Natural and Contaminant Substances, Kluwer Academic, Dordrecht.

Lambert, G., Ardouin, B., and Sanak, J., 1990, Atmospheric transport of trace elements toward Antarctica, TelIus 42B, 76-82.

Langner, J., and Rodhe, H., 1991, A global three-dimensional model of the tropospheric sulfur cycle, J. Atmos. Chem. 13,225-263.

Legrand, M., and Kirchner, S., 1988, Polar atmospheric circulation and chemistry of recent (1957- 1983) South Polar precipitation, Geophys. Res. Lett. 15,879-882.

Legrand, M. R., and Kirchner, S., 1990, Origins and variations of nitrate in South Polar precipitation, J. Geophys. Res. 95, 3493-3507.

Legrand, M., Feiet-Saigne, C., Saltzman, E. S., Germain, C., Barko~, N. I., and Petrov, V. N., 1991, Ice- core record of oceanic emissions of dimethylsulphide during the last climate cycle, Nature 350, 144-146.

Levy, H. II, and Moxim, W. J., 1989, Simulated global distribution and deposition of reactive nitrogen emitted by fossil-fuel combustion, Tellus 41B, 256-271.

Minikin, A , and Wagenbach, D., 1990, The biogenic sulfur fraction of' the coastal Antarctic aerosol, paper presented at the Seventh International Symposium of the Commission on Atmospheric Chemistry and Global Pollution (CACGP), Chemistry of the Global Atmosphere, Chamrousse, France, 5-11 September 1990.

Mulvaney, R., Pasteur, E. C., Peel, D. A., Saltzman, E. S., and Whung, E-Y., t992, The ratio of MSA to non-sea-salt sulphate in Antarctic peninsula ice cores, Tellus 44B, 295-303.

Nguyen, B. C., Mihalopoulos, N., and Belviso, S., 1990, Seasonal variation of atmospheric dimethylsulfide at Amsterdam Island in the southern Indian Ocean, J. Atmos. Chem. 11, 123-141.

Penner, J. E., Atherton, C. S., Dignon, J., Ghan, S. J., and Walton, J. J., 1991, Tropospheric nitrogen: a three-dimensional study of sources, distributions, and deposition, J. Geophys. Res. 96, 959-990.

Petit, J.-R., Briat, M., and Royer, A., 1981, Ice age aerosol content from East Antarctic ice core samples and past wind strength, Nature 293,391-394.

Polian, G., Lambert, G., Ardouin, B., and Jegou, A , 1986, Long-range transport of continental radon in Subantarctic areas, Tellus 38B, 178-189.

Poole, L. R., Jones, R. L., Kurylo, M. J., Wahner, A., Calvert, J. G., Fried, A., Hampson, R. IF., Hoff- man, D. J., Keyser, L. E, Kolb, C. E., Leu, M.:I:, Molina, M. J., Pitts, M. C., Ravishankara, A. R., Thomason, L. W., Tolbert, M. A., and Worsnop, D. tL, 1991, Chapter 3. Heterogeneous processes: laborato13~, field, and modeling studies, in R. T. Watson and D. L. Albritton (eds), Scientific Assess- rnent of Ozone Depletion: 1991, pp. 3-1-3-22.

Prospero, J. M., Savoie, D. L., Saltzman, E., and Larsen, R., 1991, Impact of oceanic sources of biogenic sulphur on sulphate aerosol concentrations at Mawson, Antarctica, Nature 350,221-223.

122 D. L, SAVOIE ET AL.

Pszenny, A. A. E, Castelle, A. J., Galloway, J. N, and Duce, R. A., 1989, A study of the sulfur cycle in the Antarctic marine boundary layer, J. Geophys. Res. 94, 9818-9830.

Pszenny, A. A. E, 1992, Particle size distributions of methanesulfonate in the tropical Pacific marine boundary layer, J. Atmos. Chem. 14,273-284.

Saltzman, E. S., Savoie, D. L., Zika, R. G., and Prospero, J. M., 1983, Methane sulfonic acid in the marine atmosphere, Z Geophys. Res. 88, 10897-10902.

Saltzman, E. S., Savoie, D. L , Prospero, J. M., and Zika, R. G., 1985, Atmospheric methanesulfonic acid and non-sea-salt sulfate at Fanning and American Samoa, Geophys. Res. Lett. 12, 437-440.

Savoie, D. L , Prospero, J. M., and Saltzman, E. S., 1989a, Non-sea-salt sulfate and nitrate in trade wind aerosols at Barbados: evidence for long-range transport, J. Geophys. Res. 94, 5069-5080.

Savoie, D. L , Prospero, J. M., Merrill, J.T., and Uematsu, M., 1989b, Nitrate in the atmospheric boundary layer of the tropical South Pacific: implications regarding sources and transport, J. Atmos. Chem. 8, 391-415, 1989.

Savoie, D. L , Prospero, J. M., Larsen, R. J., and Saltzman, E. S., 1992, Nitrogen and sulfur species in aerosols at Mawson, Antarctica, and their relationship to natural radionuclides, J. Atmos. Chem. 14, 181-204.

Savoie, D. L., Arimoto, R., Prospero, J. M, Duce, R. A., Graustein, W. C., Turekian, K. K., Galloway, J. N., and Keene, W. C., 1993a, Oceanic and anthropogenic contributions to nonsea-salt sulfate in the marine boundary layer over the North Atlantic Ocean, Z Geophys. Res. (in review).

Savoie, D. L., Prospero, J. M., Arimoto, R., and Duce, R. A , 1993b, Nonsea-salt sulfate and methane- suffonate at American Samoa, J. Geophys. Res. (in review).

Streten, N. A., 1963, Some observations of Antarctic katabatic winds, Australian Meteorol. Mag. 42, 1-23.

Tuncel, G., Aras, N. K., and ZoUer, W. H., 1989, Temporal variations and sources of elements in the South Pole atmosphere, 1. Nonenriched and moderately enriched elements, Z Geophys. Res. 94, 13025-13038.

Wagenbach, D., Gtdach, U., Moser, K., and Miinnich, K. O., 1988, Coastal Antarctic aerosol: the seasonal pattern of its chemical composition and radionuclide content, Tellus 40B, 426-436.

Whung, R-Y., 1991, A study of methanesulfonic acid in ice cores, Phi) dissertation, Univ. of Miami, Miami, Fla.

Zhu, X. R., Prospero, J. M., Savoie, D. L., Huang, E, and Huang, T., 1989, Particle size distributions off nitrate, non-sea-salt sulfate, methanesulfonate and sea-salt at Barbados, Eos, Trans. AGU 70, 1020.

Zhu, X. R., Prospero, J. M., Millero, E J., Savoie, D. L., and Brass, G. W., 1992, The solubility of ferric ion in marine mineral aerosol solutions at ambient relative humidities, Mar. Chem. 38, 91-107..

nitrogen and sulfur species in antarctic aerosols at ...dust.ess.uci.edu/ppr/ppr_spl93.pdf ·...

Documents