The contributions of snow, fog, and dry deposition to the summer flux

Transcription

University of New Hampshire
University of New Hampshire Scholars' Repository
Earth Sciences Scholarship
Earth Sciences
8-20-1995
The contributions of snow, fog, and dry deposition
to the summer flux of anions and cations at
Summit, Greenland
M H. Bergin
Carnegie Mellon University
J L. Jaffrezo
Domaine Universitaire, Grenoble, Franc
C Davidson
Jack E. Dibb
University of New Hampshire, [email protected]
S N. Pandis
See next page for additional authors
Follow this and additional works at: http://scholars.unh.edu/earthsci_facpub
Part of the Atmospheric Sciences Commons
Recommended Citation
Bergin, M. H., J.-L. Jaffrezo, C. I. Davidson, J. E. Dibb, S. N. Pandis, R. Hillamo, W. Maenhaut, H. D. Kuhns, and T. Makela (1995), The
contributions of snow, fog, and dry deposition to the summer flux of anions and cations at Summit, Greenland, J. Geophys. Res.,
100(D8), 16275–16288, doi:10.1029/95JD01267.
This Article is brought to you for free and open access by the Earth Sciences at University of New Hampshire Scholars' Repository. It has been accepted
for inclusion in Earth Sciences Scholarship by an authorized administrator of University of New Hampshire Scholars' Repository. For more
information, please contact [email protected].
Authors
M H. Bergin, J L. Jaffrezo, C Davidson, Jack E. Dibb, S N. Pandis, R Hillamo, W Maenhaut, H D. Kuhns, and T
Makela
This article is available at University of New Hampshire Scholars' Repository: http://scholars.unh.edu/earthsci_facpub/222
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 100,NO. D8, PAGES 16,275-16,288,AUGUST 20, 1995
The contributions of snow, fog, and dry depositionto the
summer flux of anions and cations at Summit, Greenland
M. H. Bergin
•, J.-L.Jaffrezo
•, C. I. Davidson
•, J.E. Dibbn,
S. N. Pandis
•, R. Hillamo•, W. Maenhaut
;, H. D. Kuhns
•, and T. Makela•
Abstract.Experiments
wereperformed
duringtheperiodMay-Julyof 1993at Summit,
Greenland.
Aerosolmasssizedistributions
aswell asdailyaverageconcentrations
of several
anionic
andcationic
species
weremeasured.
Drydeposition
velocities
forSO,:'wereestimated
usingsurrogate
surfaces(symmetricairfoils)aswell asimpactordata.Real-time
concentrations
of particlesgreaterthan0.5 gm andgreaterthan0.01 gm weremeasured.
Snow
andfog samplesfrom nearlyall of theeventsoccurringduringthefield seasonwerecollected.
Filtersampler
results
indicate
thatSO,:'is thedominant
aerosol
anionspecies,
with Na+,
NH,+,andCa•+being
thedominant
cations.
Impactor
results
indicate
thatMSAandSO,:'have
similarmasssizedistributions.
Furthermore,
MSAandSO,:'havemassin boththe
accumulation
andcoarsemodes.A limitednumberof samples
for NH•+ indicatethatit exists
in theaccumulation
mode.Na, K, Mg, andCa existprimarilyin thecoarsemode.Dry
deposition
velocitiesestimated
fromimpactorsamples
anda theoryfor drydeposition
to snow
rangefrom0.017cm/s+/- 0.011cm/sforNHn
+ to0.110cm/s+/- 0.021cm/sforCa.SOn:'dry
deposition
velocityestimates
usingairfoilsarein therange0.023 cm/sto 0.062 cm/s,asmuch
as60%greaterthanvaluescalculated
usingtheairbornesizedistribution
data.Therough
agreement
betweentheairfoilandimpactor-estimated
drydeposition
velocitiessuggests
that
theairfoilsmaybe usedto approximate
thedrydeposition
to thesnowsurface.
Laserparticle
counter(LPC) resultsshowthatparticles> 0.5 gm in diameterefficientlyserveasnucleito
formfogdroplets.
Condensation
nuclei(CN) measurements
indicatethatparticles
< 0.5 gm
arenotasgreatlyaffectedby fog.Furthermore,
impactormeasurements
suggest
thatfrom50%
to80%oftheaerosol
SOn:'
serves
asnucleiforfogdroplets.
Snowdeposition
isthedominant
mechanism
transporting
chemicals
to theice sheet.ForNO•',a species
thatapparently
exists
primarilyin thegasphase
asHNO,(g),93%of theseasonal
inventory
(massof a deposited
chemical
species
perunitareaduringtheseason)
is dueto snowdeposition,
whichsuggests
efficientscavenging
of HNO3(g)by snowflakes.
Thecontribution
of snowdeposition
to the
seasonal
inventories
of aerosols
rangesfrom45% for MSA to 76%for NHn+.Thecontribution
of fogtotheseasonal
inventories
ranges
from13%forNa+andCa•+to26%and32%forSO••'
andMSA. Thedrydeposition
contribution
to theseasonal
inventories
of theaerosolspecies
is
aslowas5% forNHn+andashighas23%forMSA.Theseasonal
inventory
estimations
donot
takeintoconsideration
thespatialvariabilitycaused
by blowinganddriftingsnow.Overall,
resultsindicatethatsnowdeposition
of chemicalspeciesis thedominantflux mechanism
duringthesummerat Summitandthatall threedeposition
processes
shouldbe considered
whenestimating
atmospheric
concentrations
basedonicecorechemicalsignals.
•Deparlment
of Civil and Environmental
Engineering, •Deparlxnent
of ChemicalEngineering,
CarnegieMellon
CarnegieMellon University,Piusburgh,Pennsylvania.
University,Pittsburgh,Pennsylvania.
2Nowat ClimateModelingand Diagnostic
Laboratory,
National Oceanic and AtmosphericAdministration,Boulder,
6Finnish
Meteorological
Institute,Air QualityDepartment,
Helsinki, Finland.
7Instituut
veer NucleaireWetenschappen,
Universityof
Colorado.
•Laboratoirede Glaciologie et
G6ophysiquede
Gent, Belgium.
l'Environnement du CRNS, Saint Martin d'Htres, France.
nInstitute
for the Studyof Earth,Oceans,and Space,
Introduction
Universityof New Hampshire,Durham.
Copyright1995by the AmericanGeophysical
Union.
Paper number 95JD01267.
0148-0227/95/95JD-01267505.00
The recentlyretrievedice coresfrom Summit, Greenland,
potentially contain an atmosphericchemical history of
approximatelythe last 250,000 years.Chemicalsignalsin ice
cores have already given insight into past climatic cycles
[Barnolaet al., 1987; Dansgaardet al., 1989; Delmasand
16,275
16,276
BERGIN ET AL.: ATMOSPHERIC
FLUXES TO SUMMIT,
GREENLAND
descriptionof the collection and extraction procedure is
Legrand,1989; Peel, 1992; Taylor et al., 1993; Mayewskiet
al., 1994]. If quantitativeestimatesof past atmospheric presentedelsewhere[Jaffrezoet al., 1994].
chemical concentrations are to be obtained from ice core data,
TSI model 3755 laserparticlecounters(LPCs) wereusedto
of particles>
it is importantto betterunderstand
the deposition
processes continuouslymeasurethe numberconcentration
0.5 grn at heightsof 20 cm and 3 m abovethe snowsurface.A
thatdeliverthesespeciesto the surfaceof glaciers.
Aerosolsandgasesaredepositedontoan ice sheetby snow, TSI model 3760 condensationnucleuscounter(CNC) was used
to monitorconcentrations
of particles> 0.01 grn (CN) at 3 m.
fog, and dry deposition.The depositionrates dependon
was used to collect
several factors in addition to atmosphericconcentrations, A TSI model 3701 multiplexer/processor
includinglocal meteorology(e.g., wind speed,temperature data and transfer it to a personal computer. Particle
inversionheight,occurrence
of fog), surfacesnowproperties concentrationdata were averagedand stored over 1-minute
(e.g.,surfaceroughness
andcharacteristics
of snowdrifts),and intervals for all instnunents.Inlets for both sample heights
were located on polyvinyl chloride (PVC) poles. The inlets
the nature of the aerosols (e.g., hygroscopicityand size
were coveredwith plastic funnelsso that snowflakesand fog
distribution)and gases(e.g., diffusivity and solubility).The
relative contributionsof the depositionprocessesto surface droplets could not readily enter the tubing. Particles were
carried to the counters via Tygon tubing (0.48 cm i.d.).
snowchemicalsignalsare still poorlyunderstood
[Jaffrezoet
Particles greater than approximately 10 !axn were not
al., 1995; Berginet al., 1995],andaccurate
estimations
of the
efficiently transported to the instruments due to inertial
necessaryparametersaffectingthe processes
are neededto
effects.A more detaileddescriptionof the samplingtechnique
predictpast atmospheric
chemicalconcentrations
basedon
concentrationsin ice cores. Unfortunately, the information is givenby Berginet al. [1994].
A 12-stageimpactordesignedat the FinnishMeteorological
available on the nature of atmosphericchemical speciesat
Summit is still incomplete.
Institute (FMI) was used to determine the mass size
distributionsof the aerosol.The impactor operatesat a flow
During the Dye 3 Gas and Aerosol SamplingProgram
(DGASP) in Greenland,year-roundsamplingof chemical
speciesin the air and snowwasconducted
[Davidsonet al.,
1993a]. Preliminary estimatesindicate that dry deposition
playsa relativelyminorrolein depositing
aerosol
species
with
a significantamountof massin thesubmicron
mode(suchas
rate of 8.33 L/min so that samplescan be obtainedin remote
regions. The impactor was calibrated at FMI for the
temperatureand pressurerange at Summit. Several impactor
sampleswere obtained during the 1993 field seasonfor ion
chromatograph(IC) analysis of major anions and cations
SO42-)
totheIceSheetwhileaccounting
forasmuchas30%of
(MSA,NO•-,SO•2-,andNH,•
+) aswellasproton-induced
X ray
emission (PIXE) analysis of minor and trace elements (W.
al., 1993a]. On the otherhand,Davidsonet al. [ 1989] reported Maenhaut et al., manuscript in preparation, 1995). The
relatively
highconcentrations
of SO42'andNO3- in fogwater impactorsubstratesare thin polycarbonatefilms coatedwith a
fine layer of Apiezon-L greasefor theIC samplesandVaseline
for Dye 3. Bergin et al. [1994] found that fog may play a
significantrole in depositingseveralchemicalspeciesto the or parafin for the PIXE samplesin order to minimize particle
ice sheetat Summit. However, a comprehensive
assessment
is
bounce.IC sampleswere extracted with 4 mL of ultrapure
currentlyunavailable
for all threedeposition
processes
on the water and analyzed in the field. PIXE samples were
Ice Sheet.
transportedto Gent, Belgium, to be analyzed at the Instituut
In thispaper,resultsobtainedduringthe 1993field season voor NucleaireWetenschappen,at Gent University.
at Summit, Greenland, are presented. First, atmospheric
concentrations
of thesoluble
ionicspecies
MSA,SO•'•-,NOs-, Snow Deposition
thedeposition
of the coarsemodecrustalspecies[Davidsonet
Na+,NH•+,K+,Mg•+,andCa•+,andmasssizedistributions
of
Fresh snow samples were collected from virtually every
thesechemicalspeciesarepresented.
Dry deposition
velocities
for the chemicalspeciesare estimatedusing impactordata snow storm from May 20 to July 8 at the GISP2 main camp.
combinedwith a theoryfor depositionto snow.Resultsusing High-densitypolyethylenetrays (36 cm x 56 cm) were usedto
surrogate
surfaces
to estimate
SO•a' drydeposition
velocities collect replicate fresh snow samples during snow events
having relatively low wind speeds(< approximately5 m/s).
The trays were mountedon PVC poles 1.5 m above the snow
surface. Sampleswere collected by a field assistantwearing
snow chemical concentrationsand inventoriesfollows. Finally,
disposablepolyethylene gloves, standing downwind of the
trays in order to minimize sample contamination.The trays
the relative contributions of snow, fog, and aerosol dry
depositionto the seasonalinventoriesof the anionsMSA, were washed several times with ultrapure water between
SO•
•',andNO3',
aswellasthecations
Na+,NH4
+,K+,andCa•+, events. During snow storms with relatively high winds the
are presentedand discussedin terms of the deposition trays did not efficiently collect fresh snow. In this case fresh
snow samples were obtained by collecting replicate fresh
processes.
surfacelayer snow with a scraperfollowing the techniqueof
Berginet al. [ 1995]. A Lexantool wasusedto scrapethe snow
Experimental Methods
samplesinto precleaned,preweighed100-mL airtight Pyrex
AtmosphericSampling
arecompared
with impactorestimates.
Next,theimpactof fog
episodeson aerosolconcentrations
is presented
with LPC,
CNC, andimpactordata.A presentation
of the fog andfresh
Daily aerosolsamplesfor analysisof major anionsand
cations were collected on Teflon Zefluor filters (1 •m pore
size) at the atmosphericcamp (ATM) located28 km SSW of
the GISP2 main camp. Samplingtook place from May 25 to
July 13. Several field blanks were collectedduring the season
and analyzed with the same method as the samples. A
bottles.Approximately10% of the freshsnowsamples(both
by number and mass) were collected in this manner. All
sampleswereweighedto determinethe freshsnowinventories
(in grams per squarecentimeter),which were then used to
determine the inventories of the soluble ionic species (in
nanogramsper squarecentimeter)for each snowevent. The
chemicalinventoryfor a speciesis the concentrationof the
FLUXES TO SUMMIT,
GREENLAND
16,277
species(in nanograms
per gram)multipliedby the freshsnow
inventoryfor the givenevent(in gramsper squarecentimeter).
Sampleswereobtainedfrom 18 snowevents.
limitsbasedon a signal-to-noise
ratioof 2 for MSA, NO3',and
Fog Deposition
injection
loopwas500uL.Thedetection
limitsforNa*,NH4*,
K+,Mg:+,andCaa+ were0.1 ppb,0.1 ppb,0.1 ppb,0.7 ppb,
The white high-densitypolyethylenetrays used to collect
fresh snow were also used to collect fog droplets.The trays
wereput out at the beginningof a fog eventandwereremoved
within 1-2 hoursof fog dissipation.Becauseof the relatively
low wind speeds(< 3 m/s) duringfogs(mostlyradiativefogs),
the samplelossesdue to wind reentrainmentwere negligible.
Replicatesampleswere collectedand weighedas previously
describedfor fresh snow. The trays were washedbetween
eventswith ultrapurewater. The fog dropletfluxes (in grams
and0.5 ppb,respectively.
per squarecentimeter)as well as inventoriesof solubleionic
speciesdue to fog deposition(in nanogramsper square
centimeter)were determined.Replicatesampleswere obtained
for 19 fog eventsfrom May 21 to July 12.
SO4
v were0.2ppb,2.0ppb,and0.5ppb,respectively.
Cation concentrations were determined using Dionex
FASTCAT 1 and 2 columns, with a switching valve. The
Aerosol filters for the anions and cations were first
extracted in clean Teflon beakers. The filters were wetted with
1 mL of low-sodium grade methanol (J.T. Baker,
Phillipsburgh,
New Jersey)followedby 14 mL of ultrapure
water. The beakerswere then hand shakenat regular intervals
for 10 min. The extractwasfiltered with a syringefitted with a
precleaned
polyethylene
filter holdercontaininga Teflon
Zefluor filter (0.5-grn pore size) to insurethat insoluble
particlesdid not enterthe IC column.Performingseveral
extractions
on singlefiltersshowedthatmorethan95% of the
anions of interest were removed using a single extraction.
Each filter was cut into two pieces to provide duplicate
samples;the concentrations
typicallyagreedwithin 10%.
Dry Deposition
Several field blanks (n = 12) were testedto determinethe
detectionlimits of the technique.The aerosolconcentration
Dry depositionflux measurements
were conducted
using
two aerodynamic
surrogate
surfaces
(symmetricairfoils).The
airfoils, developed at Carnegie-Mellon University, are
specificallydesignedto avoidformationof turbulentwakes
overthedeposition
surface.Theyare30.5 cm in diameterand
mean
oftheblanks
forMSA,NO3',
andSO•
2' were0.4ng/m
•,
5.3ng/m
•, 3 ng/m
•, andforNa+,NH4
+,K+,Mg2+,andCa•+,3.0
ng/m
•, 25 ng/m
3, 2.0 ng/m
•, 3.0 ng/m
•, and3.6 ng/m
•,
respectively,
foratypical
volume
of 70m• ofair.
manufacturedout of aluminum with a Teflon coating [Wu et
al., 1992a].The samplingtookplaceon top of 13-cm-diameter
Teflon sheets,securedto the airfoils with Teflon O-Rings.The
detectionlimits based on 3 standarddeviations (30) above the
Impactorsamples
wereanalyzedbothwith IC andPIXE.
For the IC analysesthe polycarbonate
collectionsubstrates
were each extractedwith 4 mL of ultrapurewater in clean
Teflon sheetswere cleanedwith methanol and ultrapure water
accuvettes.Several field blanks were analyzed (n = 10) to
and secured on the airfoils under a laminar
determine the detection limits. The detection limits for the
flow hood.
anionsanalyzedby IC (basedon 30 abovetheblankvalues)
Replicatesampleswere obtainedfor each samplingperiod
using airfoils positioned1.5 m above the snow surface.
Airfoilsweretransported
to andfrom the samplelocationin
high-density
polyethylene
boxesthat were doublebaggedto
minimizesamplecontamination.
The sampleswere extracted
in cleanpolyethylene
bags.For eachsampletwo extractions
wereperformed,eachwith 3 mL of ultrapurewater.Several
field blanks were also preparedand analyzed.Due to the
ng/stage,25 ng/stage,2.0 ng/stage,3.0 ng/stage,and 3.6
ng/stage,respectively.The PIXE analyseswere done as
previously
described
byMaenhautet al. [1981]. Thedetection
limitsfor PIXE analysisfor Na, K, Ca, andMg, were 27.6
ng/stage,0.8 ng/stage,1.1 ng/stage,and 7.8 ng/stage,
detection limits of the technique and relatively low
respectively.
forMSA,NO3',andSO•
•- were1.9ng/stage,
1.1ng/stage,
and
10.5ng/stage,
andfor Na+, NH•+, K+, Mg•+,andCa•*, 3.0
atmospheric
concentrations,
onlySO42-dry deposition
could
be measured.
Results
and Discussion
SeasonalInventory
Aerosol
Concentrations
Seasonalsnow inventorymeasurements
were made at the
endof thefield season,on July 11, by takingsnowcoresdown
Figures 1 and 2 show atmosphericconcentrationsof soluble
ionic speciesfrom filter measurementsat ATM during the
1993 field seasonfrom May 25 to July 13. Figure 1 showsthe
resultsfor the anionaerosolspecies.The geometricmeansand
to a plasticmesh(1 m x 1 m) setout at thebeginning
of the
field season,on May 20. A polycarbonate
tube (7.6 cm i.d.)
geometric
standard
deviations
for SO4
•', NO•',andMSA are
fitted with a high-density
polyethylene
plungerwas usedto
retrieve the snow cores (n = 3), which were subsequently 402 ng/m' (1.8), 26 ngJm'(2.9), and 5.6 ngJm'(2.2),
respectively.
SO•2' is alwaysthe dominantanion.MSA
weighed
in orderto estimate
theseasonal
snowinventory.
concentrationsare similar to valuesreportedfor the same time
Analysis
periods
duringprevious
fieldseasons,
whiletheSO••' values
are significantly higher [Jaffrezo et al., 1994; Dibb et al.,
Fresh snow, fog, dry deposition,aerosol,and impactor
1992]. TheSO•2'geometric
meanvalueis dominated
by the
samples
wereanalyzed
forMSA,NO•',SO•
•', Na*,NH•*,K*,
extremelyhigh concentrations
on the 10-dayperiodfrom May
The anionanalyses
wereperformed
usinga DionexPAX-100
column.A 26-min samplerun wasusedwith a gradientof
sodium
hydroxide
anda constant
concentration
of methanol
in
the eluant.A 700 uL injectionloop was used.The detection
have been observedduring spring and summer in past field
of relatively
highSO••-concentrations
Mg2+,andCa•+usinga Dionexseries
4500ionchromatograph.30 to June8. Episodes
seasonson the Greenland Ice Sheet [Davidson et al., 1993c]
and have been attributed to increasedbiogenic contributions
from the oceans occurring in summer [Li and Barrie, 1993;
16,278
BERGIN ET AL.' ATMOSPHERIC
FLUXES TO SUMMIT,
GREENLAND
160
--o-NO.,'
140
+ SO42' 1000
1200
120
800
100
600
8060
o
400
40
0
200
20
0
•,.,
:', : : : : : ', :', : : : : ß
0
3O
MSA
25
o
15
E
=
10
o
o
5
==
Date
Figure l. Atmosphericconcentrations
of solubleanionicspecies(concentrations
of zero indicatethat the levels
are below detectionlimits).
Jaffrezoet al., 1994], as well as to anthropogenic
sourcesfrom
North America and Eurasia[Davidsonet al., 1993b], mainly in
Spring.AerosolNO•-concentrations
duringthisperiodare also
much higher than later in the field season.Measurementsof
HNO•(g) carriedout later in the field seasonshowthat aerosol
NO•'is insignificantcompared
to HNO•(g) [Dibbet al., 1994].
Although HNO•(g) measurementswere not made for the
period of exceptionally high aerosol concentrations,it is
possible that during this time aerosol NO•-contributes
significantlyto total atmosphericnitratelevels.
Figure 2 showsthe daily aerosolconcentrations
for cation
The geometricmeanand standarddeviationof the anion
excess
fortheaerosol
samples
is 3.2nanoequivalents/m
• (2.2).
Thissuggests
thatthesummeraerosolat Summitis acidic.It
is likely thatthe acidicnatureof theaerosolis responsible
for
therelativelyhighgasphaseconcentrations
of HNO•, formic
acid,andaceticacidreportedfor severalintensivertm periods
[Dibb et al., 1994]. The acidic nature of the aerosolalso
suggests
that ammoniumnitrate aerosolgenerallydoesnot
existduringthe summerat Summit.
Specific investigations
are currentlyunderway,which
respectively.The cationconcentrations
are also dominatedby
highvaluesfrom May 30 to June8. The concentrations
of Na+
includeair massbacktrajectories,
in orderto betterunderstand
the conditionsleadingto suchepisodesof high concentrations
of chemicalspeciesfrom varioussources.
Thesestudieswill
be usedto betterunderstand
the seasonality
of the atmospheric
chemicalsignalsat Summit,andwill allow an assessment
of
therepresentativity
of our datato otherclimaticandseasonal
andCa•+ aresignificantly
greaterthanthevaluesreported
at
conditions.
species.
The geometric
meansandstandard
deviations
for Na+,
+
NH4, K+,Mg2+,andCa•+,are14ng/m
• (3.1),69 ng/m
• (1.9),
3.4 ng/m• (2.5), 6.8 ng/m• (2.1), and 29 ng/m
• (2.5),
Summit by Mosher et al. [1993] (geometricmean valuesfor
Na+ and Ca•+ of 3.2 ng/m•, and 7.0 ng/m
•, respectively).
However,Mosher et al. [ 1993] reportsdatafor the time period
from June 5, 1990 to August 27, 1990 which may not have
been affectedby the late springpeak influencingthe chemical
concentrationspresentedhere. Indeed, the concentrationsfor
the 1993 field seasonare within the range of concentrations
reportedfor a 2-week period in April at Dye 3 [Davidsonet
al., 1993c].
Aerosol Mass Size Distributions
Sulfuranalyzed
byPIXE(n = 21)andSO4
•-analyzed
byIC
(n = 11) both have bimodalmasssize distributions.
The
fractionof total sulfur in the coarsemode (greaterthan 1
rangesfrom 15% to 60% with a meanof 34% for PIXE
samples
andrangesfrom13%to 64%witha meanof 36%for
IC samples.Directcomparison
of the two setsof datais not
FLUXES TO SUMMIT,
GREENLAND
200
180
-o- Na'
--"--NH,'
160
•
16,279
Ca"
140
120
100
80
6o
4o
20
0
ß
::
....
'
25
+
o
o
i!mlgl.l,I
Mg:'
,!-I'.
:::::::::::::::::::::
Date
Figure2. Atmospheric
concentrations
of soluble
cationic
species
(concentrations
of zeroindicate
thatthelevels
are below detectionlimits).
possiblesincethe sampleswerecollectedon differentdays
masssize distributions,shownin Figure4, exhibit peaksin the
with the same device. However, the similar size distributions
accumulation
modeonly. The NI-I,+is mostlikely associated
indicates
thatthesulfurprobably
exists
asSO,:'•
Figure3 shows
MSAandSO,:' masssizedistributions
for
with the more acidic accumulation mode aerosol.
two of the four samplesanalyzedfor bothspecies.The results
Filter samplerresults show a strong correlationbetween
NOs andCa•+ and,alongwithimpactor
results,
suggest
that
thesechemicalspeciesexist togetherin the coarsemodeof the
aerosolat Summit, as alreadyshownfor midlatitudesites [Wu
themajorityof themassresidesin theaccumulation
modesin
and Okada, 1994], Previousstudiesof nitrate aerosolsin urban
the casesin Figure 3. Bimodalmasssize distributions
for
MSA andSO•a' havebeenreported
for coastalandmarine areashaveshownthatNO( lossfrom filters andimpactorscan
occur[Appelet al., 1981; Zhang and McMurry, 1992]. It is
locations[Saltzmanet al., 1983; Pszenny,1992; Quinnet al.,
thatthecoarse
modeCa• aerosol
collected
on
1993].MSA masssizedistributions
overmarinelocationstend alsopossible
indicatebimodal size distributionsfor both species,although
coastalareas.This is attributedto the greater availability of
filtersadsorbsHNO•(g) duringsampling.Sincethereis not a
greatdealof literatureavailableon the adsorption
of nitrateby
basic coarse mode sea-salt aerosols over ocean locations
calcium aerosol for the conditions at Summit, it is difficult to
to havea greaterfractionof massin the coarsemodethanat
estimatetheextentof thecoarsemodeNOs' artifact.
[Quinnet al., 1993].The masssizedistributions
of Figure3,
Overall, these data are in good agreementwith previous
with approximately
10%of the totalMSA massin the coarse
mode, could indicatea larger influenceof dry depositionas
well as of precipitationscavengingon the coarse mode
resultsand show the ability of this new impactor to resolve
sizedistributionsof many chemicalspeciesfor shortsampling
aerosolsduringtransportto the Ice Sheet[Jaffrezoet al.,
times in the Arctic.
1993].
Figure4 shows
typicalimpactor
masssizedistributions
for
Dry Deposition
severalotherchemicalspecies.Resultsfor Na, K, Ca, andMg
generallyagreewith distributions
reportedby Hillarnoet al.
[1993]for earlyspringsamples
at Dye3; withthemajorityof
themassof theseelements
beingin thecoarse
mode.TheNI-I,+
In this sectionthe dry depositionvelocitiesof the chemical
speciesare estimatedfrom the impactordata. Also, the dry
deposition
velocityestimates
for SO,a' are compared
with
16,280
BERGIN ET AL.' ATMOSPHERIC FLUXES TO SUMMIT, GREENLAND
5/23, 13:30- 5/25, 15:00
MSA
. . . SO,•
o
o
O.Ol
0.1
I
lOO
10
Aerodynamic
Diameter,
dp (um)
6128, 12:40 - 6/30, 08:40
MSA
. . . SO,ß
o
15
.<
lO
o
O.Ol
0.1
I
10
Aerodynamic
Diameter,
dp (urn)
Figure3. Masssizedistributions
of MSAandSO4:'
fortwoseparate
sampling
periods.
valuesestimatedfrom airfoil experiments.The dry deposition (possiblysaltssuchas (NH4):SO
4 ), will experience
growth in
velocitieswill thenbe used,alongwith the daily meanaerosol the viscoussublayer.The effect of this particle growth on the
concentrations,
to estimatethe daily dry depositionfluxes for
dry depositionvelocityof variouschemicalspeciesis currently
the season.
being studied(M.H. Bergin et al., manuscriptin preparation,
The dry depositionvelocity(Vd) of a chemicalspeciescan 1995). The speciesgenerallyassociatedwith the aerosolcoarse
be calculatedusingthe theoreticaldepositionvelocityto snow mode(NOs',Na, K, Mg, andCa) haverelativelyhigh average
as a functionof particlesize [Ibrahirnet al., 1983], alongwith
depositionvelocities,ranging from 0.064 cm/s for K to 0.110
mass size distributions obtained from impactor runs, as forCa.MSA,SO4:-,
andS havemassbothin theaccumulation
follows [Davidsonet al., 1993a; Hillarno et al., 1993]:
and coarse modes and the average depositionvelocities are
similarandrangefrom0.021cm/sto 0.024cm/s.NH4+ exists
'•p,
max
vd(d
p)m(dp
)ddp
:p,
max
m(dp
)ddp
Vd= p,min
(1)
p, min
wheredpis theparticle
diameter
(in micrometers),
vd(dp)is
the dry deposition velocity as a function of particle size
only in the accumulationmode and has a low mean deposition
velocity of 0.017 cm/s. The values for sulfate are within a
factorof 2 of the few datapoints obtainedover a snowsurface
by Ibrahirn et al. [1983] in the validationof their model.
Thedeposition
velocityof SO4:'
canalsobe independently
estimatedfrom airfoil data. The airfoils measurethe total dry
deposition
fluxof SO4:overa sampling
period.
Aspreviously
mentioned,due the low atmosphericconcentrationsof the
other chemical species as well as relatively high detection
[Ibrahim
et al., 1983](in centimeters
persecond),
andm(dp) limits,onlytheSO4:-fluxwasaccurately
measured
withthe
is the massconcentrationdistributionfunction(in nanograms
airfoils.During a samplingperiod,the depositionvelocitycan
per cubic centimeterper micrometer). Note that Vd in (1)
be estimated as:
corresponds
to the overall dry depositionvelocity obtainedby
integrating
vd(dp)overthemassdistribution
of a chemical
species.Table 1 gives dry depositionvelocities for several
chemical speciesusing our impactor data. The theoretical
depositionvelocities for (1) are calculatedassumingno
F
chemical species. Although it is likely that many of the
whereVd is the804:ødrydeposition
velocity(in centimeters
per second),
F is the SOl flux (in nanograms
per square
centimeter
persecond),
andC is themeanatmospheric
SO4:'
accumulationmode particles,like those containingNH4+
concentration(in nanogramsper cubiccentimeter).
particlegrowthin the surfacesnowlaminarsublayerfor all
FLUXES TO SUMMIT,
•
NO,'
-
GREENLAND
16,281
K
,•
., • i'"•
o
........
- ....
•,
o.1
O.Ol
.
---
I
lO
o
lOO
O.Ol
El)
NH4+
1
lO
Ca
El)
4O
o
.
•0
!
!
10
0
o
0.01
0.1
1
o.1
I
10
100
0.01
0.1
1
10
100
o.01
0.1
I
10
100
3O
E
•=
•0
15
lO
o
o.01
lO
100
Aerodynamic
Diameter,
dp (urn)
Aerodynamic
Diameter,
dp (urn)
Figure 4. Masssizedistributions
for severalchemicalspecies.Samplingtimesare as follows:for Na, K, Ca, and
Mg; 37 hoursfromMay 30 to June2 (dashedline), and59 hoursfromJune5 to June7 (solidline); for NO3' 15
hoursfrom June3 to June4 (dashedline), and33 hoursfrom June8 to June9 (solidline); for NH[ 12.5 hours
duringJuly 19 (dashedline), and9 hoursduringJuly20 (solidline).
Table2 showstheSO4
2' deposition
velocityresultsfrom
eight airfoil experiments conductedover the season. The
standarddeviationsfor the duplicateflux samplesare typically
from 10%-20%
of the mean values. The arithmetic
mean and
standard deviation are 0.041 cm/s +/- 0.012 cm/s. These values
are in agreementwith the depositionvelocitiesof 0.034 cm/s
and 0.021 cm/s for greasedTeflon collection sheetsused in
two exposureperiodswith the sameairfoils during the 1992
field season
at Summit[Berginet al., 1994].Althoughthedata
set is limited, similar depositionvelocitiesfor greasedand
tlngreased surfaces suggest that particle bounce and
reentrainmentare not a major influence.This may have to do
withthelackof supermicron
SOft'particles
atSummit.
Dry deposition
velocities
for SO42'
calculated
with the
impactor measurementsare systematicallylower than the
values measured with the airfoils. Comparison of
simultaneous measurements made on June 3 and June 4 show
airfoil depositionvelocitiesas much as 60% greater.This
reasonably
closeagreementfor the casesof two fairly different
typesof surfacesindicatesthat the airfoil measurements
may
beusedto approximate
drydeposition
velocities
of aerosols
to
snowat Summit,assuming
themodelof Ibrahimet al. [1983]
is an accuratedescription
of actualdry deposition
processes.
However, the differences between the two sets of results
may be due to differencesin the boundarylayer resistances
over the modeled
snow surface versus the airfoil.
This
last
flow is not well developedand it is possiblethat the viscous
sublayeris relativelysmall comparedto that in the modeled
snow surface, therefore resulting in a greater boundary layer
resistanceto snow. Another possiblefactor influencing the
results
wouldbethatSO42existsonparticles
greater
than15
grn, whichd6positon the airfoil, but are removedby the
impactorinlet and are not takeninto accountin the model.
However,the lack of particlesseenon the uppertwo stagesof
impactor
runsfordayswithout
fogsuggest
thatSO4
•' particles
largerthan 15 grn do not dominatethe dry depositionflux.
Finally, it is possiblethat turbulentburstsoccur,despitethe
airfoildesign,whicharenot considered
in themodel,thatcan
enhanceactualdry deposition[Wu et al., 1992b].
It is assumedthat atmospheric
SOadoesnot contributeto
surface
snowSO4'-'
inventories.
SOamaydeposit
onthesnow
surface
toformSO42'
byreaction
withchemical
species
suchas
I-laO
• [ConklinandBales,1993].A limitednumberof mist
chambermeasurementsmade during the field seasonsuggest
that SO• accountsfor approximately1/3 of the total
atmosphericsulfur, althoughrestfitswere highly variable
[Dibb et al., 1994]. For sea-levelArctic locations,20%-90%
of the airbornesulfur duringthe winter monthsmay exist as
16,282
BERGIN ET AL.: ATMOSPHERIC FLUXES TO SUMMIT, GREENLAND
Table 1. Dry Deposition
VelocitiesCalculated
From ImpactorRuns
Vd , cm/s
Species
Mean
Standard
Deviation
IC analysis
MSA (n = 4)
0.024
0.023
SO,,:'
(n= 11)
0.021
0.017
NO3' (n = 4)
0.072
0.053
NH4+ (n = 2)
0.017
0.011
PIXE analysis
s (n = 21)
0.022
0.016
Na (n = 5)
0.067
0.015
K (n = 5)
0.064
0.017
Ca (n = 5)
0.110
0.021
M• (n = 5)
0.078
0.016
SO: [BarrieandHoff, 1985;Li andBarrie, 1993].Duringthe
summermonths this value decreasesdue to more favorable
oxidationconditionsandis probablyabout20%-30%,mainly
from marine biogenic production. The daily SO:
eoncenlxations for the entire field season at Summit
are not
known, but the concenlxations
are probablylower than at sea
level locationsdue to oxidationand fog scavengingover the
Canada[Sirois and Barrie, 1988]. Although,this locationis
not coveredwith snow for the entire year, and experiences
much different atmosphericconcenlxations
of aerosolsand
gases as well as meteorologicalconditions,and surface
properties.Indeed, the overall air-snowinteractionsof NOy
speciesat Summit are still poorlyunderstood,specificallythe
direction of the fluxes at the snow surface
[Silvente and
Legrand,1995]. Consequently,
the datapresentedhereinclude
to surface
snowSO4:'concenlxations
dueto dry deposition exclusivelythe depositionof particulatenilxate.
cannotbe ruled out, but is considerednegligible.
In muchthe sameway thedry deposition
of HNO3(g)is not Effect of Fog on Aerosol Concentrations
ice sheetduringtransport.
Therefore,the contribution
of SO:
taken into account for the estimation of inventories. It has been
Figure5a showsLPC resultsfor 20 cm and 3 m abovethe
fraction to the total nilxate deposition over southeastern snow surface from 1200 LT on June 2 to 1200 LT on June 3. A
found that HNO•(g) dry depositioncontributesa significant
Table
2. SO4
2'DryDeposition
Velocities
from
Airfoils
,
Vd , cm/s
Start
Stop
Mean
May 31, 1993, 1100
May 31, 1993,2300
0.031
0.002
June 1, 1993, 1100
June 1, 1993, 2130
0.045
0.002
June 2, 1993, 1130
June 2, 1993, 2130
0.048
0.012
June 3, 1993, 1130
June 3, 1993, 2130
0.023
June 4, 1993, 1100
June 4, 1993, 2130
0.046
0.003
June 7, 1993, 1130
June 7, 1993, 2100
0.031
0.012
June 14, 1993, 0930
June 14, 1993, 2100
June22, 1993, 1030
June22, 1993, 2045
Mean is 0.04 1, and the standarddeviation is 0.012.
Standard Deviation
0.003
0.062
0.008
BERGINET AL.' ATMOSPHERICFLUXESTO SUMMIT, GREENLAND
16,283
a)
1..5
O..5
0
June-2, 12:00
6
//•
12
/18
Fog
•,ginsTime(hr) Fog
Ends
24
June
3,12:00
3111
100
I
I
I
I
24
June-2,
12.'00
Fog
Begins
Time(hr)
Fog Ends
June-3, 12.-oo
Figure
5. (a)LPC(particles
>0.5!.tm)
and
(b)CNC(>0.01!.tm)
results
before,
during,
and
after
afogevent.
radiative
fog eventoccurs
duringtheeveningandlastsfor particles,which are efficientlycollectedby the inlet
couldgrowto between
0.5 p.mand5.0 p.m.As
approximately
8 hours.
Theconcentration
of aerosol
particles configuration,
the
fog
dissipates,
the
concentrations
of particles> 0.5 pm
> 0.5 pm at 3 m decreases
nearlyto zeroduringthefog.
slowlyincrease,for two possiblereasons.
First,particlesare
released
asfog dropletsevaporate.
Second,
thereis a mixing
with
air
aloft
containing
fresh
particles.
The relative
al., 1992].Estimates
of theparticle
losses
in thesample
tubing
importance
of
these
two
processes
is
unknown.
The general
basedon sedimentation
indicatesamplingefficienciesgreater
patterns
shown
in
Figures
5a
and
5b
are
typical
of those
than99% for particles
from0.5 pxnto 1.0 p.m,with the
Previousstudiesat Summithave shownthat the fog droplets
havemassmeandiametersin therange15 to 20 pxn[Boryset
efficiency
dropping
to 25%at 5 pro,6% at 15pan,andless
than1%at20 Ixm.Therefore,
duetotheirlargesize,mostof
thefog droplets
arelostin thesampling
inletandtubing
upstream
of theLPC.Thelargedecrease
in thenumber
of the
largeparticles
of theaerosol
indicates
thatessentially
all of
observedfor thefog eventsduringthefield season.
Figure
6 shows
impactor
mass
sizedistributions
forSO•'
(byIC) andS (byPIXE)obtained
fromimpactor
runsforclear
daysfollowed
by evenings
withfog.Duringtheday,bimodal
size distributions are seen, with modes similar to the
reportedfor sulfurat Dye 3 by Hillamoet al.
theseparticlesgrow to form fog droplets.Previous distributions
measurements
showed
dropletconcentrations
ranging
from30
cm-3 to 165 cm'3 [Boryset al., 1992],muchlargerthanthe
concentration
of particles
> 0.5 pxnshownin Figure5a.
Particlessmallerthanthissizemustthereforebe nucleatedfor
[1993]. Duringthe eveningsas the fogs occur,the coarse
particle
modeaswellasa portion
of theaccumulation
particle
modeserveas fog nuclei,in agreement
with LPC andCNC
results.
Figures6a and6b showthatdifferentfractions
of the
modeparticlesserveasfognuclei,probablydue
the conditions
of supersaturation
seenat Summit.Indeed, accumulation
to differences
in the supersaturations
achieved
for the distinct
Figure
5bshows
thattheconcentrations
of CN aredepleted,
fogevents.
Foursetsof samples
wereobtained
similarto those
with the number concentrationof CN dropping from
approximately
200cm'3to50cm'3.These
smallparticles
are shown in Figure 6. The ratios of the aerosolmass
duringfog andduringthe daytimebeforefog
notasgreatly
affected
by fogasthelargerones,sincegreater concentrations
occurs
range
from
0.2 to 0.5. Thissuggests
thatapproximately
supersaturations
areneeded
to growsmaller
particles
[Hinds,
1982; Hiinel, 1987].
LPC results at 20 cm show an initial increase in
concentration
after fog creationfollowedby a decrease
to
valuesnear zero. The initial increaseis possiblydue to the
50% to 80% of the aerosol sulfur mass is nucleated to form
fogdroplets
duringfog events.Thelargefog dropletsarenot
sampled
bytheimpactor
either,whichhasaninletdesigned
to
exclude
particles
greaterthanapproximately
15 pxn.However,
thata fraction
of thefogdroplets
arelessthan15
greater
supersaturation
achieved
closeto thesnowsurface, it ispossible
sincethetemperature
at theground
is lowerthanat the3-m Ixm [Pandiset al., 1990] and that someare effectively
height,whichactivates
smaller
particles
at 20 cm. These collected.Also,the impactorsamplesmay includedroplets<
16,284
FLUXES TO SUMMIT,
GREENLAND
a)
,
,
I
, ,
I
mI
'q I
•'
: ,,,,,,:!:::',
o
O.Ol
I
Im
Night
(tog)
I
I I
I I
I
I
I I
I
:
•
. . ..
,I , -...... , I : • • •:
o.1
1
lO
lO0
Aerodynamic
Diameter,
dp (urn)
b)
S(PIXE)
i
i
i
i
i
Day
Night
(fog)
I
i
i
i
lO
o,
O.Ol
--
•
I
•a
II
i
I
I I '
I
I
I
!11
o.1
,
!'
I
lO
1•
Aerodynamic
Diameter,
dp (um)
Figure 6. Mass size distributionsanalyzedby a) IC for sulfateand b) PIXE for sulfurfor two separateclear days
followed by eveningswith fog.
15 pan,sampledduringthe growthandevaporation
stagesof
fog. Our estimationsof the fog scavengingefficienciesare
thereforelower limits. Theseresultsare in generalagreement
with fog scavengingefficienciesreportedat the Po valley,
Italy [Nooneet al., 1992].
SeasonalChemical SpeciesInventories
seasonat Summit (J.-L. Jaffrezo, unpublisheddata, 1992),
with the exceptionof NOs-,whichis approximately
twice as
large in the 1993 samples.The 1993 data representsfresh
snow eventswhile the 1992 data is for daily surfacesnow
samples.
The daily surfacesnowsamplesarefor layerswhich
oftenareexposed
at thesurfacefor severaldays,andtransport
of HNO•(g)out of the surfacesnowlayersmay occur[Baleset
al., 1995], resulting in lower surface snow NOs'
In this sectionthe relativecontributions
of snow,fog and
concentrations.The snow deposition inventory samples
dry deposition to surface snow chemical inventories are obtainedfrom the trayslikely representa lowerlimit sinceit is
estimated.
The chemicalinventories
for snowandfog events probable that the trays undercollect snow due to wind
are estimatedby multiplyingthe snow (or fog) inventory(in re,entrainment.The seasonalsnowinventorymeasuredat the
nanograrnsper square centimeter) for an event by the mesh
is 2.09g/cm
2+/- 0.20g/cm
a,whilethetraysaccount
for
concentration
of thechemicalspecies(in nanograms
per gram) about1.1g/cm
a.Thissuggests
thatthetraysaccount
forabout
in snow(or fog). Thedry deposition
daily fluxesarecalculated 60% of the seasonalsnowdeposition.Althoughthe meshes
by multiplying daily atmosphericconcentrations
of chemical mayhavebeenaffected
by driftingsnow,the2.09g/cm
asnow
species,shown in Figures 1 and 2, by depositionvelocity inventory is considereda more likely estimate of snow
values from Table 1.
deposition.The snow accumulationduring the 1993 field
The geometricmean and standarddeviation for the snow
season, based on the mesh measurements, is similar to the
eventinventories
is 0.054 g/cm2 (3.0), with a totalseasonal
accumulation measured for the 1989-1991
snowinventoryfrom the traysand freshsurfacesnowsamples
(which representsless than 10% of the mass inventories)of
100-stakeaccumulation
network at Summit [Bergin et al.,
field seasons at a
995].
1.23g/cm
2(1.7).Thegeometric
means
andgeometric
standard Thegeometric
meanandstandard
deviation
of thefogwater
is 0.012g/cm• (1.7),whichagrees
withthe
deviations
for the concentrations
of MSA, NOs-,SO42-,
Na+, eventinventories
NH4*,K*andCa•*insnoware2.8ng/g(2.2),435ng/g(2.1),92 fog water inventoriesreported for the 1992 field season
ng/g(2.0), 10.4ng/g(2.9), 25.8 ng/g (2.1), 2.1 ng/g(3.9), and
12.6 ng/g (2.5), respectively.The values agree with the
concentrations
measured
in surfacesnowduringthe 1992field
[Berginet al., 1994]. The geometricmeansand geometric
standard
deviations
fortheconcentrations
ofMSA,NOs-,
SO,a-,
Na+,NI-I,+, K+, andCa• in fog are2.8 ng/g(1.5),423 ng/g
FLUXES TO SUMMIT,
GREENLAND
16,285
Table 3. Seasonal
ChemicalInventoryContributions
dueto FreshSnow,Fog, andDry Deposition
Seasonal
Inventory
MSA
NO,'
SO,2-
Na+
NH,*
K*
Ca
Fresh snow
Geometric
Mean,ng/crn
•
1.1
1036
178
17
60
5.6
46
Geometric Standard deviation
1.7
1.1
1.1
1.4
1.1
1.2
1.1
%, Mean
%, StandardDeviation
45
13
93
1
61
5
Fog
71
8
76
4
69
6
63
3
Geometric
Mean,ng/cm:
0.8
70
76
3.0
14
1.4
8.3
Geometric Standard deviation
%, Mean
%, Standard Deviation
1.2
32
9
1.1
6
1
1.1
26
3
1.6
13
6
1.3
19
4
1.5
19
6
1.4
13
4
Dry
Geometric
Mean,ng/cm
•
0.6
7.9
34
3.8
4.4
1.0
12
Geometric Standarddeviation
%, Mean
%, Standard Deviation
1.3
23
7
1.2
1
1
1.2
13
6
1.1
16
4
1.1
5
1
1.2
12
3
1.1
18
2
(2.2), 429 ng/g (2.2), 17.6 ng/g (2.5), 74 ng/g (1.9), 7.3 ng/g
(2.0), and 50 ng/g (2.6), respectively. Concentrationsof
chemical speciesin fog from a limited number of samples
collectedduring the 1992 season(n = 3) have valueswithin the
rangereportedfor the 1993 season.
Table 3 showsthe seasonalinventory contributionsdue to
+
species
(Ca•*, K* andNa*), the fog anddry deposition
inventoriesare generallysimilar, but in this caseit is due to
theirrelativelyhigh dry depositionvelocities.
Figure
7 shows
theSOn
•'andCa•* inventories
ona day-byday basis,from May 21 to July 13 for snow,fog, andaerosol
dry deposition.
For bothSOn
•' andCa•*, snowdeposition
snow,
foganddrydeposition
forMSA,NOs-,
SOn
•-,Na+,NHn,
accounts
for thethreehighestdailyinventories.
Dry deposition
K+, andCa•+.The seasonal
chemicalinventoriesfor snoware
and fog depositioncontributeequal fractionsto the seasonal
estimatedby adding the chemical inventoriesfor each snow
event. The chemicalinventoryfor each snow event, collected
with the trays, is estimated by assuming that the trays
collectedonly 60% of the total snowinventoryfor the event.
The uncertaintyin chemical snow depositioninventoriesare
accountedfor by consideringthe variability in the measured
snow inventoriesas well as analytical error. The geometric
standarddeviation (GSD) in the seasonalfog inventory is
basedon the variability in the chemical speciesconcentrations
and fog massinventoriesof replicatesamples.The uncertainty
in the seasonaldry depositioninventoriesare determinedusing
the dry depositionvelocity standarddeviationsin Table 1.
For all chemical species,snow depositionis the dominant
processcontributingto the seasonalinventory.For NO(, a
chemical speciesthat exists primarily in the gas phase,up to
93% of the seasonalinventory is due to snow deposition,
althoughdry deposition
of HNO•(g) is not considered
hereand
may contribute to surface snow chemical inventories. This
suggestsefficient scavengingof HNO•(g) by snow crystals
[Silvente and Legrand, 1995]. It is also possible that an
Ca•* inventory,
whileforSOn'-'
fogdeposition
dominates
dry
deposition.
The dry deposition
inventories
aregreatest
during
the periodsof high atmosphericconcentrations.
Overall, the
inventoriesare driven by the atmosphericconcentrations,
tendingto be highduringtheperiodof May 30 to June8.
The chemicalinventories
of snow,fog, anddry deposition
presentedaboverepresentvaluesfor the summermonths.The
relative contributionsof these processesto chemical
inventories
duringdifferenttimesof theyear,or fromyearto
year, may vary, dependingon the changesin the parameters
that control the fluxes. The snow depositionchemical
inventories
for a giveneventdependon the snowinventory
and atmosphericchemical concentrationsas well as other
factors that include snow crystal size, shape, altitude of
formation,extentof timing, and concentrationand the aerosol
masssize distributions[Bell and Saunders,1991; Wang,
1992]. The dry depositioninventoriesdepend on the
atmosphericaerosolconcentrations
and masssize distribution,
as well as wind speed, atmosphericstability, and
characteristics
of the surfacesnow.The fog inventoryfor an
aerosolfor an eventdepends
on the atmospheric
. additional
NOvspecies
existsthatcontributes
to NOs-measured atmospheric
in surface
snow[Dibbet al., 1994].ForSOn
2'andNHn
+,which concentrationof the chemicalspeciesas well as factorsthat
have significantmassfractionsin the accumulationmode, the
includethedurationof the fog event,fog height,fog droplet
seasonalfog inventory is greater than the dry deposition
inventory.AlthoughMSA also has a significantfractionof
mass in the accumulationmode, dry depositionand fog
depositioncontributeapproximatelythe sameto the seasonal
inventory,due to the relativelylow concentrations
of MSA
duringfogeventsduringthefield season.
Indeed,onlyonefog
event occursduring the relatively high MSA concentration
episodefrom June26 to July 4. For the coarsemode aerosol
settlingvelocity, and fraction of the initial aerosolmass,R,
incorporated
into fog droplets.The largerparticlesare more
readilyactivated
to formfogdroplets.
TheR valuesfor SOn'-'
predicated
from impactorsamplesrangefrom 0.5 to 0.8, andit
is likely that coarsemode aerosolchemicalspecies,suchas
Ca•*, haveR valuescloserto 1.
Thusseasonalvariationsin atmospheric
concentrations
as
well as severalparametersaffectingthesefluxes to the snow
16,286
BERGIN ET AL.' ATMOSPHERIC FLUXES TO SUMMIT, GREENLAND
a)
19.3
17.6
12
20.8
41.5
•'•Fog
9
6
0
:
:; : :::.l
Date
b)
.7.7
9.3
.6.2
3
[-•Fog
2
I Dry
Dep.
Date
Figure7. (a) SO,•'and(b)Ca•*inventories
forsnow,fog,and•
deposition.
surface must be measured if accurate estimates of the
The dry depositionvelocity estimatedfrom impactormass
contributions
of the depositionprocesses
to the annualsignals size distributionsfor NH•, an accumulation
modeaerosol,is
of chemical species in snow are to be made. Year-round 0.017cm/s+/- 0.011cm/s. ForMSA andSO•
2',whichhave
samplingof aerosolchemicalspeciesandchemistry
of fog and mass in both the accumulationand coarsemodes, the dry
snow sampleson an event basis are required, supplemented depositionvelocitiesare 0.024 cm/s +/- 0.020 cm/s, and 0.021
with properdevelopmentof modelsfor the determinationof cm/s +/- 0.017 cm/s, respectively.The coarsemode aerosol
the seasonalvariabilityof theprocesses.
speciesNa, K, Mg, and Ca have dry depositionvelocitiesof
+
0.067 cm/s +/- 0.015 cm/s, 0.064 cm/s +/- 0.017 cm/s, 0.078
cm/s +/- 0.016 cm/s, and 0.110 cm/s +/- 0.021 cm/s,
Conclusion
respectively.
Estimates
of theSO•:' drydeposition
velocity
usingairfoils
samplers
indicatethatSO42'
is themainaerosol
anion,while are as much as 60% greater than the values found using
Experimentsperformed at Summit, Greenlandwith filter
the cation concentrations
are generallydominatedby Na+,
NH4*,andCa:*.Impactor
results
indicate
thatMSAandSO•
2'
have similar mass size distributions.Impactor results also
impactor size distributionsand the dry depositionmodel to
snow of Ibrahim et al. [1983]. This is possibly due to
differencesin the boundarylayer resistancesof the airfoil and
snowsurfaces.
Also,growth
of aerosol
SO•
:' in the
showthatSO••' existsin the coarsemodeas well as the modeled
accumulation mode. Although only a limited number of
surfacesnow viscoussublayeris not considered.Overall, the
+
impactorrunsare availablefor NH• (n = 2), resultsindicate rough agreementbetweenthe airfoil and impactor-estimated
that this speciesexistsmainly in the accumulationmode. Na,
K, Mg, andCa exist,primarilyin thecoarseparticlemode.
Laserparticlecounters(LPCs) indicatethat particles> 0.5
gm serve as nuclei to form fog droplets.A large fraction of
thesedropletssettleto the snowsurface.Condensation
nucleus
counter(CNC) resultssuggestthat particlessmallerthan 0.5
gm are not as greatly affected by fog as larger particles.
dry depositionvelocitiessuggests
that the airfoilsmay be used
to approximatethe dry depositionto the snowsurface.
Theseasonal
inventories
of MSA, NOs-,SO4:',
Na+, NH•+,
K*, andCa:*aredominated
by snowdeposition.
ForNOs',a
chemicalspeciesthat existsprimarily in the gasphase,93% of
the seasonalinventoryis due to snow deposition,suggesting
efficientscavenging
of HNO•(g) by snowflakes,
althoughonly
Impactor
results
indicate
thatSO••' in thecoarse
modeserves thedrydeposition
of particulate
NOs'is considered.
ForSO•
:'
asfognucleiandthatasmuchas 80% of theSO42'
aerosol aswell asNH•, the fog inventoryis significantlygreaterthan
+
massis incorporatedinto fog droplets.
the dry deposition inventory, due to relatively low dry
BERGIN ET AL.: ATMOSPHERIC FLUXES TO SUMMIT, GREENLAND
16,287
deposition
velocities.The contributions
of fog and dry Davidson,C. I., J. R. Harfington,M. J. Stephenson,M. J.
Small, F. P. Boscoe,and R. E. Ganalley,Seasonalvariations
deposition
to theMSA seasonal
inventory
aresimilar,dueto
in sulfate, nitrate, and chloride in the Greenland Ice Sheet:
the relativelylow atmospheric
MSA concentrations
during
relationto atmospheric
concentrations,
Atmos.Environ.,23,
periods
of theseason
havingfogevents.Dry deposition
and
2483-2983, 1989.
fogdeposition
contribute
similarlyto theseasonal
inventories Davidson, C.I., et al., Chemical constituents in the air and
of thecoarse
modespecies
Na*,K+, andCa•+ dueto their
relatively
highdrydeposition
velocities.
In orderto accurately
estimatethe contributionsof the flux processesto the annual
inventory,
year-round
sampling
of snow,fog,andatmospheric
chemical concentrations are needed.
Theresultssuggest
thatall threedeposition
processes
are
snow at Dye 3, Greenland, I, seasonalvariations,Atmos.
Environ., 27A, 2709-2722, 1993a.
Davidson, C. I., J. L. Jaffrezo, M. J. Small, P. W. Summers,P.
M. Olson, and R. D. Borys, Trajectory analysisof source
regionsinfluencingthe southGreenlandIce Sheetduring
the Dye 3 Gas and Aerosol SamplingProgram,Atmos.
Environ., 27A, 2739-2750, 1993b.
Davidson, C. I., et al., Chemical constituentsin the air and
important
in determining
theannual
inventories
of aerosols
at
Summit.
Theprocesses
of foganddrydeposition
mayplayan
snowat Dye 3, Greenland,II, Analysisof episodes
in April
evenmoreimportant
roleduringdryperiods,
forexample
ice
1989, Atmos.Environ., 27A, 2723-2738, 1993c.
ages.
Theresults
pointouttheimportance
ofunderstanding
the
parameters
whichgovernthe deposition
processesand Delmas, R.J., and M. Legrand, Long-term changesin the
concentrations
of major chemicalcompounds(solubleand
developing
methods
to estimate
theseparameters
so that
insoluble) along deep ice cores, in The Environmental
chemical
signals
in ice corescanbe usedto determine
past
Record in Glaciers and Ice Sheets, Dahlem Workshop
atmospheric
concentrations.
Acknowledgments. We would like to thank Matt Penderfor
collectionof field samplesand SallieWhitlow for adviceconcerningIC
sampleanalysisas well as JanCafmeyerfor technicalassistance.
We
also thank GISP2 SMO, PICO, and the 109th NYANG, as well as the
Danish ResearchCommissionfor grantingpermissionto study in
Greenland.This work wassupported
by NSF grantDPP-9123082.The
PIXE analysiswas supported
by the Belgian"NationaalFondsvoor
Wetenschappelijk
Onderzoek"and from the Belgian"GlobalChange"
programs.
References
Report, edited by H. Oeschgerand C. C. Langway Jr.,
JohnWiley and Sons,New York, 1989.
Dibb, J. E., J. L. Jaffrezo, and M. Legrand,Initial findingsof
recentinvestigationsof air-snow relationships in the
Summitregionof the GreenlandIce Sheet,J. AtmosChem.,
14, 167-180, 1992.
Dibb, .l.E., R. W. Talbot, and M. H. Bergin, Soluble acidic
speciesat Summit,Greenland,Geophys.Res.Lett., 21(15),
1627-1630, 1994.
H'finel, G., The role of aerosol properties during the
condensational
stageof cloud:a reinvestigationof numerics
and microphysics,Contrib. Atmos. Phys., 60, 321-339,
1987.
Hillarno, R. E., W. Maenhaut, J. L. Jaffrezo, S. Balachandran,
Appel,B. R., Y. Tokiwa,andM. Haik,Sampling
of nitrates
in
ambientair, AtmosEnviron.,15,283-289, 1981.
Bales,R., Nitric acidin firn: discussion,
in Ice CoreStudiesof
Global Biogeochernical
Cycles, edited by R. Delmas,
NATO-ASI
Ser.Springer
Verlag,NewYork,in press,1995.
Bamola,J. M., D. Raynaud,Y. S. Korotkevich,
andC. Lorius,
Vostokicecoreprovides 160,000-year record of
atmospheric
CO2,Nature,329,408-414,1987.
Barfie, L. A., and R. M. Hoff, Five years of air chemistry
observationsin the Canadian Arctic, Atmos. Environ., 19,
1995-2010, 1985.
Bell, D. A., and P. R. Saunders,The scavengingof high
altitudeaerosols
by ice crystals,Atmos.Environ.,25A,
801-808, 1991.
Bergin,
M. H., J.L. Jaffrezo,
C. I. Davidson,
R. Caldow,
andJ.
E. Dibb, Fluxesof chemicalspeciesto the GreenlandIce
Sheetat Summit by fog and dry deposition,Geochern.
Cosrnochim.Acta, 58, 3207-3215, 1994.
C. I. Davidson, and V. M. Kerminen, Atmospherictrace
elementsat Dye 3, Greenland 1, Size distributionand dry
depositionvelocities, Atmos. Environ., 27A, 2787-2802,
1993.
Hinds,W. C., AerosolTechnology,JohnWiley and Sons,New
York, 1982.
Ibrahim, M., L. A. Barfie, and F. Fanaki, An experimental
investigationof the dry depositionof particles to snow,
pine trees, and artificial collectors,Atmos.Environ., 17,
781-788.
1983.
Jaffrezo, J. L., R. E. Hillarno, C. I. Davidson, and W.
Maenhaut, Size distributionsof atmospherictrace elements
at Dye 3, Greenland-H, Sources and transport,Atmos.
Environ., 27A, 2703-2708. 1993.
Jaffrezo .l.L., C. I. Davidson,M. Legrand, and J. E. Dibb,
Sulfate and MSA
in the air and snow on the Greenland
Ice
Sheet,J. Geophys.Res.,99, 1241-1254, 1994.
Jaffrezo, J.L., J.E. Dibb, R.C. Bales, and A. Neftel, Current
statusof atmospheric studies as Summit (Greenland)
and implicationsfor futureresearch,in Ice Core Studiesof
Global BiogeochernicalCycles, edited by R. Delmas,
of GlobalBiogeochernical
Cycles,editedby R. Delmas,
NATO-ASISer. SpringerVerlag,New York, in press,1995.
NATO-ASI
Ser.Springer
Verlag,NewYork,in press,1995.
Borys,R. D., D. Del Vecchio,
J.L. Jaffrezo,
J. E. Dibb,andD. Li, S. M., and L. A. Barfie, Biogenic aerosolsin the Arctic
troposphere,
1, Contributionsto total sulfate,J. Geophys.
L. Mitchell, Field observations, measurements and
Bergin,M. H., C. I. Davidson,
J. L. Jaffrezo,J. E. Dibb,R.
Hillarno,H. D. Kuhns,andT. Makela, in Ice Core Studies
preliminary
results
froma studyof wetdeposition
processes Res., 98, 20,613-20,622, 1993.
influencing
snowandice chemistry
at Summit,Greenland, Maenhaut,W., A. Selen,P. Van Espen,R. Van Grieken, andJ.
W. Winchester,PIXE analysisof aerosolsamplescollected
in Precipitation,Scavenging
and Atmospheric
Surface
over the Atlantic Ocean from a sailboat, Nucl. lnstrum.
Exchange,
editedby S. E. Swchartz
andW. G. N. Slinn,pp.
Methods, 181,399-405, 1981.
1693-1702,Hemisphere
Pub.,Philadelphia,
1992.
Conklin,M. H., and R. Bales,SO2 uptakeon ice spheres: Mayewski, P. A., et al., Record drilling depth struck in
Greenland,EoS Trans. AGU, 75(10), 113, 1994.
liquidnatureof theice-airinterface,
J. Geophys.
Res.,98,
16,851-16,855, 1993.
Dansgaard,
W., J. W. C. White,andS. J. Johnsen,
Theabrupt
terminationof the YoungerDryas climateevent,Nature,
339, 532-534, 1989.
Mosher, B. W., P. Winklet, and J. L. Jaffrezo, Seasonal trend
in aerosolchemistryat Dye 3, Greenland,Atmos.Environ.,
27A, 2761-2772, 1993.
Noone, K. J, et al., Changes in aerosol size-and phase
16,288
FLUXES TO SUMMIT,
distributions
dueto physicalandchemicalprocesses
in fog,
Tellus, 44B, 489-504, 1992.
Pandis, S. N., J. H. Seinfeld, and C. Pilinis, Chemical
compositiondifferencesin fog and cloud dropletsof
different sizes,Arm.
os. Environ., 24A, 1957-1969, 1990.
Peel,D., Merelythetip of theice core,Nature,359, 274-275,
1992.
Pszenny,
A., Particlesizedistributions
of methanesulfonate
in
thetropicalPacificmarineboundary
layer,J. Atrnos.
Chern.,
14, 273-284,1992.
GREENLAND
Exchange,editedby S. E. SwchartzandW. G, N. Slinn,pp.
87-96, HemispherePub.,Philadelphia,1992.
Wu, P.M., andK. Okada,Natureof coarsenitrateparticlesin
theatmosphere-a
singleparticleapproach,
Atmos.
Environ.,
28, 2053-2060, 1994.
Wu, Y. L., C. I. Davidson, D. A. Dolske, and S. I. Sherwood,
Dry depositionof atmosphericcontaminants:
the relative
importanceof aerodynamic,boundarylayer, and surface
resistances,Aerosol Sci. Technol., 16, 65-81, 1992a.
Wu, Y. L., C. I. Davidson, and A. G. Russell, A stochastic
model for particledepositionand bounceoff,Aerosol Sci.
Quinn,P. K., D. S. Covert,T. S. Bates,V. N. Kapustin,D.C.
Technol., 17, 231-234, 1992b.
Ramsey-Bell,and L. M. McInnes, Dimethylsulfide/cloud
condensation
nuclei/climatesystem:Relevantsize-resolved Zhang, X., and P. H. McMurry, Evaporativelossesof fine
particulatenitratesduringsampling,Atmos.Environ.,26A,
measurementsof the chemical and physicalpropertiesof
3305-3312, 1992.
atmospheric
aerosolparticles,J. Geophys.
Res.,98, 10,41110,427, 1993.
Saltzman,E. S., D. L. Savoie, R. G. Zika, and J. M. Prospero,
Methane sulfonic acid in the marine atmosphere, J.
Geophys.Res.,88, 10,897-10,902,1983.
Silvente,E., andM. Legrand,A preliminarystudyof the airsnowrelationshipfor nitric acid in Greenland,in Ice Core
Studiesof Global Biogeochernical
Cycles, edited by R.
Delmas, NATO-ASI Set. SpringerVerlag, New York, in
press,1995.
Sirois, A., and L. A. Banfie, An estimateof the importanceof
dry depositionas a pathwayof acidicsubstances
from the
atmosphere
to thebiosphere
in easternCanada,Tellus,40B,
59-8O, 1988.
Taylor,K. C., G. W. Lamorey,G. A. Doyle,R. B. Alley, P. M
Grootes,P. A. Mayewski,J. W. White, and L. K. Barlow,
The 'flickering switch' of late Pleistoceneclimatechange,
Nature, 361,432-436,
1993.
Wang,P. K., Comparison
betweenthe collectionefficiencyof
aerosol particles by water drops and ice crystals, in
Precipitation, Scavenging and Atmospheric Surface
M. H. Bergin,NOAA, R/E/CG1, 325 Broadway,Boulder,
CO, 80303.
J.-L. Jaffrezo, LGGE du CRNS, Domaine Universitaire,
Rue Mollere, BP96, 38402, Saint Martin d'H6res, France.
C. I. Davidson,and H. D. Kuhns,Departmentof Civil and
EnvironmentalEngineering,Carnegie Mellon University,
PGH, PA, 15213.
J. E. Dibb, IEOS, Universityof New Hampshire,Durham,
NH, 03824.
S. N. Pandis, Department of Chemical Engineering,
CarnegieMellonUniversity,PGH, PA, 15213.
R. Hillarno, and T. Makela, Finnish Meteorological
Institute,Air Quality Department,Sahaajankatu
22 E., SF00810, Helsinki, Finland.
(ReceivedOctober21, 1994; revisedJanuary13, 1995;
acceptedFebruary20, 1995.)

The contributions of snow, fog, and dry deposition to the summer flux

Transcription

Similar documents