JOURNAL
OF GEOPHYSICAL
Role of mineral
aerosol
RESEARCH,
VOL. 101, NO. D17, PAGES 22,869-22,889, OCTOBER
as a reactive
20, 1996
surface
in the global troposphere
Frank J. Dentener,1,: Gregory R. Carmichael,3 Yang Zhang,3,4Jos Lelieveld,: and
Paul J. Crutzen s
Abstract. A global three-dimensionalmodel of the troposphereis used to simulatethe
sources,abundances,and sinksof mineral aerosoland the speciesinvolvedin the
photochemicaloxidant, nitrogen, and sulfur cycles.Although the calculatedheterogeneous
removal rates on mineral aerosolare highly uncertain,mainly due to poorly known
heterogeneousreaction rates, the reaction of SO2 on calcium-richmineral aerosolis likely
to play an important role downwindof arid sourceregions.This is especiallyimportant for
regionsin Asia, which are important and increasingemitters of sulfur compounds.Our
resultsindicate that the assumptionthat sulfate aerosolfollows an accumulationmode size
distribution,is particularlyin Asia likely to overestimatethe sulfate aerosolclimatecoolingeffect.An even larger fraction of gasphasenitric acid may be associatedwith and
neutralizedby mineral aerosol.Interactionsof N205, 03, and HO2-radicalswith dust are
calculatedto affectthe photochemicaloxidantcycle,causingozone decreasesup to 10% in
and nearbythe dust sourceareas.Comparisonof theseresultswith limited available
measurementsindicatesthat the proposedreactionscan indeed take place, althoughdue
to a lack of measurementsa rigorousevaluationis not possibleat this time.
1.
Introduction
Mineral aerosolsproducedfrom windblownsoilsare an important componentof the earth-atmospheresystem.It is estimated that 1000 to 3000 Tg of such aerosolsare emitted annually into the atmosphere[Jonaset al., 1995;d'Almeidaet al.,
1987]. In comparison,the global estimateof secondaryaerosols(e.g., carbonaceoussubstances,
organics,sulfate, and ni-
trate) is -400 Tg yr-• [Preining,
1991;Tegenet al., 1996].
Furthermore, the emissionsof mineral aerosols may be increasingsubstantiallyas the arid and semi-arid areas expand
due to shifting precipitation patterns and land use changes
associatedwith overgrazing, erosion, land salinization and
mining/industrialactivities[Sheehy,1992]. Dust stormshave
become a distinct feature in many regions around the globe,
includingeast Asia, west Africa, and South America [Schultz,
1979;Prosperoet al., 1979].
The effectsof mineral aerosol in the atmosphereremain
largelyunquantified.It is knownthat atmosphericaerosolscan
influence radiative transfer by absorptionand scatteringof
solar and terrestrial radiation, and by changing the optical
propertiesof cloudsthroughmodificationof the distributionof
rent estimatessuggestthat the climate forcing due to aerosols
linked to fossilfuel combustionand biomassburning largely
offsetthat due to greenhousegasesin portionsof the tropics
and industrialized areas [Kiehl and Briegleb,1993]. Mineral
aerosols should exert a similar
influence.
The radiative
effects
of mineral aerosolsare under study;for example, Tegenand
Fung [1994, 1995], Tegenet al., [1996],Liet al. [1996],Andreae
[1996]; however,large uncertaintiesremain due to the lack of
detailed information on size distribution, chemical composition, surface properties, source strengths,and atmospheric
transport and removal processes.
Aerosols also play important roles in many biogeochemical
cycles,by providing reaction sites and serving as carriers for
many condensedand sorbedspecies.For example,Luria and
Sievering[1991]haveclaimedthat the heterogeneous
oxidation
of SO2 on aerosolsmay accountfor nearly 60% of the oxida-
tion of SO2to SO42-in the marinetroposphere.
In addition,
the chemicalconversionof SO2 to sulfate on sea salt particles
hasbeen shownto be a significantsourcefor nonseasalt (nss)
sulfate in marine boundary layers [Chameidesand Stelson,
1992; Sieveringet al., 1992]. Aerosols and clouds may also
impact the photochemicaloxidant cycle.Dentenetand Crutzen
cloudcondensation
nuclei(CCN) [Charlsonet al., 1992].Cur- [1993] have shownthat reactionsinvolvingN205 and NO3 with
sulfate aerosolsappreciably alter the troposphericlevels of
NO,,,
03 and OH, and Lelieveldand Crutzen [1990] assessed
•Department
of Air Quality,Wageningen
University,
Wageningen,
Netherlands.
the effectsof cloudson troposphericphotochemistry.Zhang et
2Institute
for MarineandAtmospheric
Research,
Utrecht,Nether- al. [1994] have investigatedthe influence of mineral aerosol
lands.
(alsoreferred to as Kosa aerosol)on the troposphericoxidant
3Department
of Chemical
andBiochemical
Engineering,
andCenter
for Global and RegionalEnvironmentalResearch,Universityof Iowa, cyclein a box model analysis.Ozone wascalculatedto decrease
due to reactionson mineral particles by 10-20%. However,
Iowa City.
4Environmental
and EnergySciences
Division,BattellePacific little is known regardingthe impact of ambient mineral aeroNorthwestLaboratories,Richland, Washington.
solson chemistryof the global tropopshere.
SMax-Planck-Institut
ffir Chemie,Mainz,Germany.
In this paper we look more closely at the role of dust.
Copyright1996 by the American GeophysicalUnion.
Specifically,we addresswhether mineral aerosolscan affectthe
Paper number 96JD01818.
chemistry
of SOx(:SO2 + sulfate),NOy (:NOx + HNO3),
0148-0227/96/96JD-01818509.00
and 03. We focus on the global aspectsof reactionsof trace
22,869
,•,o •u
DENTENER
ET AL.. MINERAL
AEROSOL
tion
So•urce
area?umber
sizedistrib•u
10--t.x
'x...............
I .......
ON A REACTIVE
Source
le+05
area
SURFACE
mass
.........
I
size distributi
........
I
'
ß
lø2
--..-%
on
' ' ')' ....
/,/
.'/
ß
e+04
,/x'•
,/•,/
ß
,
/
//
ß /
\
F løo
//
ß
o
,'
• 1e+05
'•,,
/
//
,
,
/
//
o
•z
• 1e+02
10-2_
/ ,,' /
_Shettle(bockground)
x,
x
-•-6/L._.
....
s•tt•(o•t•to•).
normal
condition
(thiswor•
I
.duststorm(this work)
1e•00
0.1
'\
"\,
/i'/
..Shettle
(duststorm)
\'\'\,
e+O
•//"'"fl
//-\'\
ln_4
L
/
/' ,////
1.0
10.0
100.0
•
le-01
r [um]
_._C)11•LLI•
_ _duststorm (this work)
•normc]l
, , ,,,,,I
1e+00
condition (this work)
........
I
le+01
i
le+02
r [um]
F!gure1. Prescribed
aerosolnumberandmasssizedistribution
duringduststormandnormalconditions,
basedon the measurements
byd•41meida
[1987]in Senegal.
For comparison
the dustdistributions
givenby
Shettle[1984] are alsopresented.
gaseson mineral aerosol.For this purposewe use a global instantaneous
windsare calculated,a statisticalapproachwas
three-dimensional(3D) coarseresolutionmodel to simulate chosento describethe dustsources.
Desertarealcoveragefor
the globalsources,transports,and removalof mineral aerosol. arid regions in Africa, Asia, Australia, South America, and
The resultsof the dustsimulations
areusedto provideaerosol North America was taken from Pye [1987]. The monthly
surfaceand massfor heterogeneousreactionsof SO2, HOx, amountof dust storm dayswas calculatedfrom Pye [1987],
NOy,and03.
Schutz[1979], and Uematsu[1983].Typically,1-3 high dust
daysper monthare simulatedin the highdustseason.In areas
2.
Model Description
for which we did not have detailed information on the seasonal
distributionof dust storm days,the yearly number of dust
stormdayswas monthlydistributedusingthe model winds.A
The global three-dimensional
transportmodel Moguntia complicatingfactor is that observationsof dust storms are
used in this work has a horizontal resolution of 10 ø x 10 ø and
reportedmostlyfor populatedregions,wherelandusechanges
10 verticallayersof 100 hPa thickness[Zimmermann,1988]. may have affectedthe dust mobilization[Tegenand Fung,
2.1. Global Transport/Chemistry Model
greatlydiffersregionTransportis described
by monthlyaveragewinds[0ort, 1983] 1995].Also,the numberof observations
and eddy diffusion coefficientsbased on the standard devia-
tionsof thosewinds.In additionto the monthlymean transport, deep cumulusconvectionis parameterizedaccordingto
Feichterand Crutzen[1990].Convectivetransport,however,is
only applied to relativelyinsolubletrace gases,whereasall
aerosolcomponents
(includingmineralaerosol)wereassumed
to be effectivelyscavengedin the cumuli.The model includes
backgroundC3-C2-CH4-CO-NOx-HO,,
photochemistry,
coupled with reducednitrogen (NHx) and sulfur (DMS, SO2,
ally.
In our model aerosol size distributions
for the dust storm
and "normal"dayswere prescribedin the surfacelayer [--•0400 m] basedon the measurements
by d9tlmeida[1987]in
Matam, Senegal(Figure 1). Matam is situatedin an alluvial
flood plain,with a relativelylargefractionof smallparticles,
whichcanbe subjectto long-rangetransportafter beinglifted
into the atmosphere.
The prescribedsizedistributions
on "normal" and"duststorm"daysare takenfromdMlmeida's[1987]
SO]-) chemistry.
Anthropogenic
emissions
of SOxandNOx "windcarryingdust"and "sandstorm"cases.Considerableunweretakenfrom GEIA (GlobalEmissions
InventoryActivity)
[Benkovitzet al., 1996].Details of the chemistrycalculations,
andresultsof photochemical,
sulfurandreducednitrogensimulations(without dust) have been presentedby Langnerand
Rodhe[1991],Dentener
andCrutzen[1993,1994].In additionto
certainties are associated with the use of these size distribu-
tions. For comparison,we also presentin Figure 1 the size
distributions
of Shettle's
[1984]dustmodel.AlthoughShettle's
and d'Almeida'sdust storm size distributionscomparewell,
our "normalcondition"hasmoremassin the largersizeranges
the previouslymentioned tracers,in this work sulfate and ni- thanShettle'sbackground
case.Thereissomeevidence[Dulac
trate associatedwith dust are explicitlytaken into consider- et al., 1992;Y. Balkanski,personalcommunication,
1996]that
ation.
Shettle'sdistributionmay be more appropriate.The uncer-
2.2.
Dust
Model
Mineral aerosolhas been modeledby Wefers[1990]and
Wefersand Jaenicke[1990]. As in the Moguntia model no
tainW in terms of aerosolsurfaceavailablefor reactionsthus
may be up to a factor of 10 for a specifiedamountof mass!
Mineral aerosolwastransportedin 10 discretesizeintervals
AFi [/xm]with
DENTENER
ET AL.' MINERAL
AEROSOL
ON A REACTIVE
SURFACE
22,871
substituteo•by 3/,which is the reaction probability or reactive
stickingcoefficient.If uptake is not rate-limited by aqueous
andr, •,•o= [0.1,0.3,0.7,1.0,3.0,7.0,10.0,30.0,70.0,100.0]brm. phasediffusionor reaction, o• = 3/,in other caseso•represents
Usingan averageaerosoldensityof 1.8 g cm-3 andthe air an upper limit for ?.
density,the number concentrationAn l' in sizeinterval zXr,was
HO,, reaction probability. Odd hydrogenradicalssuch as
convertedto massmixingratio whichis the quantityneededin HO: and OH are likely to react on atmosphericdustparticles,
the transportroutine.
probablythrough catalyticpathwaysinvolvingredox reactions
Aerosol particles are removed by wet and dry deposition. with iron or copper [Rossand Noone, 1991]. A studyby MatFollowingthe parameterisationbyJungeand Gustafson[1957], thijsenand Sedlak [1995] showedthat in cloud droplets,in the
thewetremoval
rateLp (s 1)istakentobeproportional
tothe presenceof iron or copper, the destructionof HO: resultsin
average
rateof formationof precipitation
P (g m 3s •) andto the formation of hydrogenperoxide through the reactions:
the dimensionless
scavengingefficiencye:
HO2+ Fe(III)-• Fe(II) +O2 + H +
(s)
At, = [0.9r,, 1.1r,]
(1)
P
HO:+ Fe(II)(+ H+)-•Fe(III) + H:O:
whereL is the liquid water contentof the precipitatingcloud,
(6)
Mozurkewiche! al. [1987]measuredthe stickingcoefficientc•of
for whichwe adopted1 g m-3. Input data for this scheme HO 2 on aqueousparticlesto be larger than 0.2. Hanson et al.
originate from measuredclimatologicalprecipitationfields at
the Earth's surface[Jaeger,1976]. The vertical distributionof
precipitationis scaledto the releaseof latent heat with height
[Newellet al., 1974].To accountfor the insolubleaerosolfraction in mineral aerosol,the removal by wet depositionwas
arbitrary assumedto be only half as effectiveasthat for soluble
aerosol(e = 0.5).
Further, particlesare removedby Stokesgravitationalsettling. Above roughly10/•m the settlingvelocityis so high that
most emitted particlesare directlydepositednear the source.
In the surfacelayer all particlesare, in additionto gravitational
settling,subjectto transportin eddiesand impactionand stick-
[1992] observed uptake coefficientsc• for HO 2 larger than
0.01-0.05 on water and H2SO4 surfaces,respectively.Hdinel
[1976] showedthat Saharan dust at RH > 50% takes up significant amountsof water. If the dust particles contain some
water, aqueousphasetrace metal redox reactionsare likely to
be very fast, and a ? value for HO 2 of 0.1 is certainlyjustified.
In this caseHO 2 uptake is essentiallydiffusionlimited. We do
not have information on the uptake of HO 2 on dry aerosol,and
in thiswork we assumedit to be equallyfast under dry asunder
wet conditions.Laboratory experimentsof HO 2 uptake on dust
particles, preferably under realistic atmospheric conditions,
would be very valuable. OH radicalsare alsolikely to be taken
ing at the surface.A constant
valueof 0.1 cm s-• wasassumed
up by dust aerosolswith reaction rates similar to HO 2. We do
for this process,being valid mostly for submicronparticles,
not take OH uptake into account,becauseeven very fast hetwhich are inefficientlyremovedby gravitationalsettling[Duce
erogeneousOH losson mineral aerosolwould be insignificant
et al., 1991].
comparedto gas phase OH destruction.In this work we have
Aerosol number, surface and mass concentrations were obneglectedthe uptake of hydrogenperoxide on dust particles,
tainedby usingRombergintegrationof aerosolconcentrations
although the accommodationcoefficientof H202 on aqueous
•Xn, [Presset al., 1986]. Resultsof the dust simulationsand
surfacesis known to be large [Worsnopet al., 1989]. Becauseof
comparisonswith measurementsare presentedin section3.1.
reaction (6) dustparticlesmay even be a sourcerather than a
sink for H202. Destruction reactions of H202, such as with
2.3.
Reaction Rates on Dust Particles
SO2, are too slow to justify high reaction probabilities. This
Uptake and reaction of gases on aerosol surfaces. The
contrastswith aqueous phase chemistry where large cloud
pseudo-first-order-rate
coefficient
k• Is 1]whichdescribes
the droplet volumes are available, so that substantialamounts of
net removalrate of gasphasespeciesjto an aerosolsurface,is H202 may react with SO2 (see section on HNO3 and SO2
given by
reaction probability).
N205 and NO 3 reaction probability. N205 is readily taken
up by aqueoussurfacesthrough the reaction:
k,-
k•,./(r)n(r) d(r)
1
(3)
N2Os+ H20•2
HNO3
(7)
with n(r) d(r) [cm-4] is the numberdensityof particlesbe- The reactionprobability3/(N205) on sulfuricacidparticleswas
tweenr (aerosolradius)and r + dr, and kd./ is the size determined by Hanson and Lovejoy [1994] to be 0.06-0.12.
dependent
masstransfercoefficient
[cm3s•] calculated
using Morzurkewichand Calvert [1988] measuredreaction probabilthe Fuchsand Sutugin[1970] interpolationequation:
ities of 0.05-0.09 on ammonium sulfate particles, and Fan
Doren
et al. [1990] measureduptake coefficientsof •0.05 on
4zrr D/V
water
and
sulfuric acid. To our knowledge no comparable
= + Xn(X+ - -)/3.)
(4) reaction probabilities
on mineral aerosol have been deterwhereDe[cm:s •] isthegasphase
molecular
diffusion
coef- mined. In this studywe use a relatively high ? of 0.1, assuming
ficientfor the gasunder consideration,Kn is the dimensionless that ? on mineral aerosolis identical to that on other particles.
•udsen number(= h/r), and h the effectivefree path of a gas Recentwork by Fentere! al. [1996], however,indicatesthat the
molecule in air, V is a ventilation coe•cient, which is close to reactionprobabilityon dry salts,may be lower by 2-3 ordersof
1, and a is the dimensionless mass accomodation coe•cient
magnitude.Thus the applicationof a ? of 0.1 may overestimate
(alsocalleduptakeor stickingcoe•cient), which is definedas the removal of N205 in the dust sourceareaswith low relative
the number of moleculesadsorbedby the surface divided by humidities(see also section3). On the other hand, N205 prothe number of collisions with the surface. In this work we
duction is most effectiveduring the night [Dentenetand Crut-
22,872
DENTENER
ET AL.: MINERAL
AEROSOL ON A REACTIVE
SURFACE
An alternative way of estimating reaction probabilities of
zen, 1993]when, at leastin the boundarylayer, relative humidity (RH) is at maximum.Thus,underthesecircumstances,
dust SO2 on dust is basedon dry depositionobservations,or more
Sehmel
particlesare likely to containsomewater and a highvalueof •/ correctlyexpressed,surfaceresistancemeasurements.
seemsjustified.
[1980] presentsdata on SO2 depositionon calcareoussoils,
The nitrate radical NO 3 may also be subjectedto reaction cements
andFe203rangingfrom0.5-3cms-•; thesemeasurewith dust aerosol [e.g., Li et al., 1993]. However, the reaction ments may, however, include aerodynamicresistances.From
mechanismis not very clear. Sensitivitystudieswere performed thesedepositionvelocities•/may be estimatedusingthe equaand we found that additional heterogeneousNO 3 radical re- tion [Schwartz,1992]:
actions did not lead to substantial additional NOx removal.
4 Vd
Therefore, in this study NO 3 reactions on dust have been
=
Vm
ignored.
HNO 3 and SO2 reaction probability. HNO3 and SO2 upUsinga molecular
speedVmof 400ms-•, wecalculate
a •/value
take coefficientson water dropletsare on the order of 0.1 [Van
for SO2rangingfrom 5 x 10-s to 3 x 10-4 for deposition
Doren et al., 1990; Worsnopet al., 1989]. It is assumedthat
velocities
Vdbetween0.5-3 cm s-•. Dependencies
of dry deuptake of HNO3 on dust particles is followed by a simple
positionon soil humidity (or relative humidity of air) have
neutralisation reaction, for example'
been measured,but were not very important for calcareous
HNO3+ CaCO3--->
Ca2+NOj+ HCOj
(8a) soils[e.g., Payrissatand Beilke, 1975]. As expected,a strong
dependencyof depositionvelocity on soil p H was observed.
HNO •,
• Ca2+NO.•
- + HCO;• --'-,•,•,l,,•,.•/z,
r'" t•xTr•• • HzOa-CO2
,
(8b) Dry depositionof gas phase HNO 3 is determined almost entirely by the aerodynamicresistance[Hanson and Lindberg,
at least when there is alkaline material available in the aerosol.
1991, and referencestherein], and surfaceresistancerc is very
For sufficientlywetted aerosol this should be a fast process, low. Assuminga surfaceresistanceof HNO 3 smallerthan 10 s
to be largerthan10-3, whichis
and probablythis reaction is even fast under dry conditions. m-1, •/(HNO3) is calculated
The situationfor SO2-uptakemay be more complicated,asthe consistentwith the laboratoryobservationspresentedearlier.
In our studywe use a T(HNO3) of 0.l, and •/(SO2)of 3 x
sorptionof SO2 is followedby an oxidationreactionof SO2 or
associatedanions.For example,Beilkeand Gravenhorst[1978] 10-4 for areaswithRH < 50% and•/(SO2)= 0.1 at RH >
present a reaction mechanismproceedingvia metal-sulfito- 50%. The latter value is based on the assumptionthat the
complexes,with HO 2 and SOj as chain carriers.In addition, oxidationof SO2 by 03 in the aqueousphaseis very fast, an
assumptiononly justified for high p H values. Therefore we
SO2 can be oxidizedby the reaction [Maahs, 1983]:
assume
that the uptakeof SO2 (and HNO3) onlytakesplaceif
2SO 3 q-O3•SO•
q-0 2
(9)
the alkalinity from the dust aerosolexceedsthe acidityfrom
the dust-associatedsulfate and nitrate. Alkalinity in soilsis to
The solubilityandthusthe reactionof S(IV) with 0 3 is strongly a great extent determined by the calcium carbonatecontent.
p H dependent,and for pH > 8, in the presenceof water, the
Calciumcontentsof soilsin the arid regionsof Asia rangefrom
oxidationof S(IV) is sufficientlyfast to make heterogeneous
4 to 8% (by weight) [Wangand Wang,1995].LojJe-Pilot
et al.
reaction of SO2 on mineral aerosol essentiallygas phase dif[1986]report a calcitecontentfor the Sahararangingfrom 5 to
fusionlimited. (As indicatedbefore, the reactionof SO2with
30% (Ca: 2-12%). Arid regionsin the United Statescontain3
H202 is not very important on dust particles, becausethe
to 8% Ca2+ [Gillette
etal., 1992].CaCO3mayneutralizeacidity
reactionrate of (12) is not very fast, and in contrastto cloud through the reaction:
droplets,not enoughwater volume is presentto reach significant conversionrates.)
CaCO3+ 2 H +--•Ca2++ CO2q-H20
( l l)
Observational data support the formation of sulfate and
an averageCa2+ contentof 5%,
nitrate on mineral aerosol. Single particle studies of Asian We usein our calculations
aerosols,usingspottestsand transmissionelectronmicroscopy which is somewhatlarger than the global averagecrustalCa
(TEM), clearlyshowthat the mineral aerosolis coatedwith contentof 3.6% given by Jaenicke[1988]. It shouldbe noted
sulfatesand nitrates [Parungoet al., 1995]. In addition, labo- that our descriptionof SO2removalasdependenton alkalinity
ratorymeasurements
indicatethat25-400 mgSO2g-• maybe and water content is rather similar to the one proposedby
removedby fly ash, cement, and dust [Mamaneand Gottlieb, Judeikeset al. [1978].
In addition to heterogeneousoxidation of SO2 on mineral
1989; Judeikiset al., 1978; Dlugi et al., 1981]. Judeikeset al.
aerosol
by ozone, in our model SO2 is oxidized in clouds
[1978] measuredreactive stickingcoefficientson various materialslikeFe203,flyash,andsoot,ranging
from10-3 to 10-6 throughreactionwith ozone (9) and H202:
From measurementsby Dlugi [1981] a similar range of values
HSO•-aq
q-H202a
q-->
HSO•-aq
q-H20
(12)
can be derived.Reactive stickingcoefficientsdecreasedduring
prolongedexposure,but muchlessso at highrelativehumidity and in the gasphaseby the OH radical:
(RH). Becauseof the nature of the dust sourceregions(hot
SO2+OH(+ O2)--•SO3+HO2
(13a)
and dry) water vapormixingratiosand RH in the vicinityof the
source regions are quite low. Thus the uptake of water by
503 q-H2O--->H2504
(13b)
aerosol particles is expectedto take place only in air masses
which are mixed with more humid air at some distance from
the latter reaction being fast. We assumethat the gas phase
the sourceregionsor at night in the boundarylayer. Outside H2SO4 producedby the latter reactionsequenceis either formthe dustsourceareas,relative humiditiesin the boundarylayer ing newparticleswhichare removedby coagulation,or directly
are generallyhigher than 50%, enablingthe SO2 oxidationby condensingon both the preexistingaccumulationrange (am0 3 or radical reactions.
moniumbi) sulfate aerosoland mineral aerosolaccordingto
DENTENER
ET AL.: MINERAL
AEROSOL
ON A REACTIVE
SURFACE
22,873
1000.0
lOOO
Nanjing
Nanjingß
•Taichung
ßJinan
Beijing•
ß Jinan
100
Tsushin2a
ßKagoshima
ß Beijing
Tokyoß
•
Chejußw
100.0
Taichung
Kangwha
Nangøkuß
•'mmah
Tsush•ma
ß Yangyang
_
ß Kangwha
Otobe•
b..•,Yangyang
Kashlmaß
ß Fukue
•ø•:yø
Ok•ß
ß Kagosh•ma
Fukue 'Otobe
Kash•maß
ß ßCheju
ß ß Tmmah
Hachijo ß
Amam• ß
10.0
Hachuøß
ß Nangoku
Ok•
ßAmam•
1.0
1
110
100
1000
nssGa
2+[neq/m
3]
.0
10.0
100.0
0.0
nssGa
2+[neq/m
3]
Figure2. Measured
ratioof (a) aerosol
nitrateand(b) nss-sulfate
to nss-calcium
(inn-equivalence
m-3) in
easternAsian aerosol.All data exceptUI data are from data collectedfrom weeklyfiltersduringJune 1992
to May 1993[Fujitaand Takahashi,1994].The UI data is from dailysamplestakenat ChejuIsland,Korea
from 1992 to 1994 [Carmichaelet al., 1995].
mass[Zhanget al., 1993], and that soil derived dust may conthuskeeptrack of sulfateproducedby gasphaseand/orcloud tain particulateorganicmatter and mineralorganiccomplexes
chemistry,and sulfate producedon mineral dust particles. in concentrationsthat depend on the soil type and land use
Mineral aerosol, coated with sulfate and nitrate, may in fact historyof the area of origin.
providea ratherefficientcloudcondensation
nucleiandcollect
Measurementsin Japan indicate that mineral aerosol assoadditionalsulfateduringcloud-processing.
The importanceof ciatedwith longrangetransport(Kosaparticles)contain2 to 3
this processhasnot yet been assessed
in this work.
timesmore organiccarbon,includinga variety of nonmethane
The implicationsof the abovechemicalmechanisms
for sul- hydrocarbonsincludingalcoholsand organicacids,than nonfate and nitrate formation on mineral aerosol should be noted.
kosaparticles[Ohta, 1991].There is indirectevidencethat O_•
For nitrate, the mechanism results in aerosolswith maximum uptakeon dustaerosolscanbe of somesignificance,
for examnitrate levels determinedby the stoichiometricrelation with ple, ozone dry depositionmeasurementson soils and sand.
calciumasdeterminedby (8). In addition,the aerosolnitrateis Aldaz [1969]givesan averagedry depositionvelocityof about
found with the same size distribution as the mineral aerosol.
0.5 cm s-1 on sandand dry grass,with dry bare soilsbeing
Thesemodel assumptions
are consistentwith observational approximatelytwice as effectivein destroyingozone as wet
data of the chemical compositionof aerosolsin east Asia. soils.Turneret al. [1973] estimatedan averagedepositionveParticulatenitrate is plotted againstnon sea salt (nss)calcium locityon soilsof 0.5 cms • but alsoindicated
thatthe actual
(on an equivalentbasis)in Figure2a. We obtaina very high surfaceresistanceon loam and organicsoilsmay be quite low
their surfaceratios. In this versionof the Moguntia model we
correlation, with the stoichiometric relation consistentwith
(8). Sizeresolvedchemicaldata [Horaiet al., 1993]showsthat
(0.2-0.4s cm-•), andmuchhigher(0.7-1.3s cm-•) for high
soil water contentsand coarsesand. Weselyet al. [1981] deternitrate is associatedwith the nss-calcium,which representsthe mined a low surfaceresistanceof 10 s m • for cold bare soils,
--1
mineral aerosolsizefraction (Figure 3). Nitrate is not present
corresponding
to a maximumdepositionvelocityof 10 cm s
in submicronaerosol,indicatingthe relativelylow solubilityof
Galballyand Roy[1980]measureda surfaceresistancefor sand
HNO3 in the acidic sulfate aerosol and the absense of
of 40-60 s m --• (1.7-2.5cm s •). The lowestsurfaceresisNH4NO 3 aerosol.
tances were observed during the day, possibly related to
In the case of sulfate, the above mechanism indicates that
changesin soil temperatureand photoactivationof surfaces
SO2reactionswith (mineral)aerosolsarep H dependent,and
sites. Furthermore, Galbally and Roy [1980] noted very low
thus sulfate should correlate with nss-calcium. However, in
contrast to nitrate, most sulfate is found in the accumulation
mode. A secondarypeak, coincidentwith nss-calciumis also
shownin Figure 2b. Consistentwith the above mechanism,
nss-sulfate
concentrations
are higherthan that predictedby the
mineral reaction alone. Size resolveddata in Asia (Figure 3)
also supportthat sulfateis found in both the accumulation
mode and on mineral aerosolsurfaces[Horai et al., 1993].The
fraction of sulfate associated with the mineral
from
aerosol varies
10 to 90%.
03 reactionprobability. To our knowledge,no directmeasurementsof O3 uptake by dust aerosol exist. Further, the
mechanismfor O3 destructionis unclear.Reactionswith trace
metals,suchas iron and manganese,and oxidationof organic
material by ozone are offered as mechanisms.It shouldbe
noted in this respectthat dust contains4 to 10% of iron by
surface
resistances
(30 s m-•) for loamandcalcareous
soilsat
low moisturecontents(<20% water content),rapidlyincreasing with increasingsoil moisture.Active sitesfor ozone destructiongradually being coveredwith H20 moleculeswas
offered as an explanation.Garland [1976], as referenced in
--1
Sehmel[1980],measureddepositionvelocitiesof 0.14 cm s
on sand,0.22cms-• on CaCO3 and0.84cms-• and1.76cm
s-• on soilscontaining
27 and4% H20, respectively.
Stockeret
al. [1987],usingthe eddycorrelationtechniques,
measuredlow
depositionvelocitieson sandand rocksin Nevada,of 0.11 cm
s-• duringthe day,fallingto 0.03cm s-• overnight.
Similar
results are obtained by Glisten et al. [1995] who measured
deposition
velocities
in the Saharaof 0.15cms-• duringthe
day and 0.065cm s-• duringthe night.Theseregionsmay,
however,represent"dead" sourceareasin termsof uplift of
22,874
DENTENER
ET AL.: MINERAL
AEROSOL ON A REACTIVE
SURFACE
7 x 10-4, perhapsdueto coverage
of surfacesitesbyoxygen-
S042-
ated groups.
The reaction systemsof NO2-soot and NO2-soil have been
more extensivelymeasured, for example, by Judeikeset al.
[1979].Uptakecoefficients
of NO2 rangedfrom 6.4 x 10-s,
3.4 x 10-s, 3 x 10-s to 1 x 10-s on sandyloam,cement,
0
0.1
1
10
30
Aerodynamicdiameter [gin]
5
NO34
F%O3, and sand, respectively.Uptake coefficientsare also
found to graduallydiminishafter multiple exposures,but surfaceswere found to be reactivatedby precipitation [Judeikes
and Wren,1978].Comparingthe uptakecoefficientsof NO2 on
soot with O3 on soot, the latter may be nearly 6 times faster
[Smithet al., 1988]. It is unknownwhether ozone destruction
on dustparticlesis comparableto destructionof O3 and NO 2
on carbonaceousmaterials, or if a parameterisationdue to
reactionwith iron is more appropriate.
All in all, the range of measuredlaboratoryuptake coefficientsseem to agree with the 3/values derived from the dep-
ositionmeasurements.
In our studywe useda 3/of 5 x 10-s,
being most consistentwith the information presentedabove.
We estimatea lower limit for direct ozone destructionby dust
3
of 1 x 10-s, andan upperlimit of 3'(03) = 2 x l0 -4, which
2-
correspondsto the highestmeasureddepositionvelocities,in
agreementwith the uptake experimentsof Fendelet al. [1995]
andStephens
et al. [1986],assuminga highiron/organicfraction
of the aerosol. Note that this upper limit is well below the
measurednondestructive
massaccomodation
coefficienta(O3)
0
01
1
10
30
Aerodynamicdiameter[gin]
Ca 2+
> 2 x 10-3 onwatersurfaces
[Utteret al., 1992],whereas
the
earlier 3'valueswere reactiveuptake coefficients.
3.
3.1.
Results
Dust Calculations and Comparison With Previous
Results
0
0.1
....
•,,,]
1
.........
1[0 30
Aerodynamicdiameter [!um]
Figure 3. Observedmasssize distributionof sulfate,nitrate,
and calcium of aerosol measured at Kagoshima, Japan in
March 1993 [Horai et al., 1993].Dashedline representsthe non
sea salt fraction.
mineral aerosol.From (10), with va rangingfrom 0.1 to 3 cm
s-•, 3'is calculated
to rangefrom3 x 10-4 to 1 x 10-s
Global dust emissions. The calculatedglobal dust source
in the model is 1800 Tg/yr for particlesr < 10 •m and 15500
Tg/yr for particlesr < 30 •m. A relativelysmallfractionof the
emitteddustis transportedout of the sourceregions,16% and
3%, for particlessmaller than 10 and 30 •m, respectively.It
shouldbe noted that the calculatedaveragedust concentrationswould be only half the present ones,if the "dust-storm"
dayswere omitted.The calculatedemissionsare in the rangeof
previousestimates,rangingfrom 200-5000 Tg/yr [Pye, 1987;
Tegenand Fung; 1994; Duce, 1995;Andreae, 1995], although
the absoluteamount of global dust emissionsis stronglydependent on the choice of the maximum dust particle size,
which is still consideredto be "emitted." Gilletteet al. [1992]
estimated a dust source for the United States of 19 Tg for
particlessmallerthan 10 •m. Our estimateof 24 Tg (r < 10
•m) for the United Statesis in reasonableagreementwith his
source,consideringthe simplificationsmade in our work.
Dust concentrations. The calculatedannual-averagedand
February-March-Aprilaveragedsurfacelayer mineral aerosol
massconcentrationsare presentedin Figures4a and 4b, respectively.The springmonthsare, at least in Asia, the "high
dust" season.The sourceregionscan be clearly distinguished,
with averageconcentrationsduring the springdust seasonbeing about twice as high comparedto the yearly average.Concentrationsvary by 4 ordersof magnitudefrom more than 300
Laboratoryuptake experimentsof 03 and NO2 on sootand
iron aerosolsgive additionalinformation on the ozoneuptake
probabilityfor the "organic"and "iron" reaction,respectively.
Fendelet al. [1995] determineduptake coefficientson average
of 4 x 10-4 on fresh carbonaceousaerosols,and deduced
similar coefficientsfor iron aerosols.Assumingthat reactions
with iron mostlydetermine the destructionof ozone, a typical
iron contentof about 5% in dustaerosol[Pye,1987;Nishikawa, to lessthan0.1 •g m-3. Dustconcentrations
decrease
rapidly
1991a,b]wouldtranslate
to 3'= 2 x 10-s. Stephens
etal. [1986] with height, and in the free tropospherezonal mean concenshowedthat the uptake coefficientsof ozoneon carbonaceous rationsrangefromnearzeroto 3 •g m-3 (Figure4c).
The concentrations
within the sourceregionsare difficultto
surfacesmay be initially high but decreaseby more than a
sincemodeleddustconcentrations
in
factor 10 after repeated exposureresultingin an average3' of compareto observations
DENTENER
,
ET AL.: MINERAL
AEROSOL ON A REACTIVE
SURFACE
, •veerlyeveragemineralaero,solmes,
s [ug/,m3] su/aee
22,875
,
60N
..
-o
5ON
(5.2
E¸
3os
6os
18ow
15ow
12ow
9ow
6ow
3ow
o
3OE
60E
90E
120E
150E
180E
Feb-¾arch-Xtpr.mintral oer,osolmo,ss [ugXm5] svrface ,
60N
30N
o
o.1
30S
60S
i
i
180W 150W 120W
90W
60w
....
lOO
30W
0
year,I)/..... ge,
30E
60E
lug/m,3] .... I ....
90E
120E
150E
180E
lOO
c
-200
2oo
300
0.1
400
-"' /
5OO
700
gOS
0j •_
•o%
o
l:
,
60S
/,/
5OS
-500
-800
,
EQ
-4oo
-700
'b 'x
ø
800
-1000
/•/•'•
-- -500
5ON
60N
1000
90N
Figure4. Calculated
surface
layermineralaerosol
concentrations
(micrograms
percubicmeter)(a) annual
average
(b)February-March-April,
(c) annualandzonalaverage.
Isolines
are0.0,0.1,0.3,1.0,3.0,10.0,25.0,
50.0, 100.0, 200.0, and 300.0.
99 8'76
DENTENER
Measured/calculated
ET AL.:
MINERAL
AEROSOL
ON A REACTIVE
dust concentrations
1000.
O0
100.000
.......
.."!•
SURFACE
mass size distribution Apri
.........
•
........
i
.........
Sal
100.00
......
'......
31
_ 10.00
[
•83
.20._3345
'1::I I
'•fi•-•'"'
'"47•'•
15
E 1.00[ 44.a•l 1
o
1.000
0.10 0 .
0.01 [.....
0.01
....
0.10
r= .
1.00
_
10.00100.0
:
0.100
1000.00
measurements
Barbados
0.010
Figure 5. Comparisonof calculatedand measureddustconcentrations downwind oI the source areas. References are t,
Shemya,Prosperoand Savoie[1989], Annual average;2, Midway, Prosperoand Savoie [1989], Annual average; 3, Oahu,
Prosperoand SavoieJ1989],Annual average;4, Fanning,Prosperoand Savoie[1989],Annual average;5, Nauru,Prosperoand
Savoie[1989], Annual average;6, American Samoa,Prospero
0.1
10.U
100.0
r [urn]
Figure
a•d Savoie
[1989],Annualaverage;
7, NewCaled.,Prosperobados.
and Savoie[1989], Annual average;8, Norfolk, Prosperoand
Savoie[1989],Annual average;9, Enetawak,Duceet al. [1980],
April; t0, Enetawak,Duce et al. [1980], Sept.; tl, Phill. Sea,
Prospero[1979], Annual average; 12, Japan, Iwasaka et al.
[1993];13, Yaku Island,Nishikawaand Kanamori[t99ta], winter/spring(Non-Kosa);14, Sal;Prospero
et al. [1979],July/Aug./
Sept.; 15, Barbados,Prosperoet al. [1979],July/Aug./Sept.,16,
Miami, Prosperoet al. [1979], July/Aug./Sept.;17, Cayenne,
Prosperoet al. [1981], Annual average; 18, Bay of Bengal,
Prospero[1979], March/April; 19, Mediterranean, Prospero
[1979], April/May; 20, Mediterranean, Prospero[1979], July/
Aug.; 21, Hawaii, Parringtonet al. [1983], spring;22, Hawaii,
Parrington et al. [1983], other seasons;23, North.Atlantic
(20N-30N), Prospero[1979], annualaverage;24, North.Atlantic2 (30-40N), Prospero[1979],June-Sept;25, North.Atlantic3
(40N-50N), Prospero[1979], May-Aug; 26, S. Atlantict (2838S),Prospero[1979],Oct.-Dec; 27, S.Atlantic2(t1-36S),Prospero [1979],Dec.-Feb; 28, Cape Town,Prospero[1979],March;
29, W. CoastN. Africa, Prospero[1979],FEb; 30, SouthPole,
Shaw [1979], annual average;For references(31)-(49) see
Duce [1995]: 31, Beijing, China; 32, Xian, China; 33, Xiamen,
China; 34, Malipo, Korea; 35, Poker Flat, Alaska; 36, Barrow,
1.0
6.
Calculated mass size distribution at Sal and Bar-
culationsis r - 0.67; measuredconcentrationsare generally
reproducedwithin a factor of 3 by the model. The dust concentrationsin Barbados, Cayenne, and Miami are strongly
underestimated(by a factorof t0), whereasthe concentration
at Sal is rather well predicted,whichmay be an indicationthat
wet removal of mineral aerosolis too strongin our model. On
the other hand,the underpredictionmay alsobe a resultof the
existenceof a transport mode above the marine boundary
layer, which is not reproducedby our model [Westphalet al.,
1987, 1988; Karyampudiand Carlson,1988]. The situationis
further complicatedby the observationsof Chiapello et al.
[1995] that Sal Island may not be a very representativesite for
the long rangetransportof dust,as the seasonalcyclesof dust
concentrations
at Sal and Barbados
are almost anticorrelated
due to different seasonaltransportpatterns.
The same may be concludedfrom the underestimationat
some remote Pacific Islands. In addition, in Asia our calculated
Alaska; 37, Thule, Greenland; 38, Ft Smith, Canada; 39, NW
sourcemay have been underestimated;for example,Kangand
Indian Ocean;40, Northern Indian Ocean;41, CanaryIslands; Sang[1991]report for Seoulan annualaveragemineralaerosol
42, Barbados; 43, Bermuda; 44, Mace Head; 45, Okushiri; 46,
concentrationtransportedfrom the Chinesecontinentof 55 txg
Hachiojima; 47, Chichijima;48, New Zealand; 49, Amsterdam
Island.
m-3 whichishigherthantheaverage
concentration
calculated
by our model [25 txgm-3]. Within the United Statesour
calculatedconcentrationscan be comparedto the coarsefraction aerosolmeasuredin the IMPROVE network[Malm et al.,
the sourceregionsrepresentthe productof averagedustcon- 1994]. Our calculateddust concentrationsoverestimatemeacentration and areal coverageof arid regionswithin the t0 ø x
sured concentrationsby a factor 3-5, which can be partly ex10ø gridboxes,while measurementsreflect high spatial and plainedby the cut-offdiameterof the measurements
beingonly
temporal variabilities.Measured and modeled dust concentra- t0 txm, whereas in the model in the source areas particles
tions on the lee-sideof the sourceregionsare easier to com- larger than t0 txmalsorepresenta significantamountof mass.
Dust size distribution.
The calculated mass size distribupare, and we focus on the aerosol fraction subject to longrange transport(roughly<t0 txm), sincethe coarseparticles tion for the month of April at Sal (located off the African
are depositedby sedimentationin the direct vicinity of the coast) and Barbadosare presentedin Figure 6. Comparing
emissions area.
with the sourcesizedistributionpresentedin Figure 1, the size
A comparison of the measured and calculated surface level distributionsnarrow toward particlessmallerthan t0 p_mdue
concentrations,predominantly at backgroundsites, removed to sedimentation,in agreementwith the calculationsof Schutz
from the dust sourceareasis presentedin Figure 5. The cor- [1979].The mineralmasstransportedout of the sourceregions
relation coefficient between the measurements and model calis dominatedby aerosolsbetween t and 5 txm,which is also
DENTENER
,
60N
ET AL.: MINERAL
,
AEROSOL
ON A REACTIVE
anqual aveFage coumn dulst
SURFACE
. ,
,
60E
90E
22,877
-
30N -
E¸-
30S
-
60S
-
180W
150W
120W
90W
60W
50W
0
50E
120E
150E
180E
Figure 7. Calculatedannualaveragecolumndust(milligramsper squaremeter).Isolinesare 0.0, 10.0,25.0,
50.0, 75.0, 100.0, 300.0, 500.0, 1000.0, 3000.0, 5000.0.
observedin measurementsof Kosa eventsin Asia [e.g., Nish- three fixed radii of 0.2, 1, and 5/•m. The sourceregionswere
ikawa et al., 1991a,b], indicatingthat our sourcesize distribu- basedon the GISS vegetationdata set, and a sourcefunction
tion can be used for areas outside of Africa. However, as dependenton wind speedto the power3. The sourcefunction,
of about
discussed
in section2.2, a large uncertaintyis associatedwith however,wasadjustedto yield averageconcentrations
the use of this source size distribution.
I /•g/m3 overthe oceans.Calculateddustdistributions
in the
Dust column abundance. Figure 7 showsthe calculated lowest model layer source regions showed somewhatlower
annual averagedust columnamounts.The valuesrange from dust concentrationsthan ours,probablydue to the neglectof
lessthan 10 to more than 300 mg m-2. Elevatedcolumn aerosol larger than 5 /•m.
amounts are found off the coast of east Asia, off the westcoast
A recent studyby Tegenand Fung [1994] utilizes the same
of Africa, and over the Indian Ocean.These featuresare qual- model, but a more advanceddescriptionof the sourceregions,
betweenclay,silt, and sandfractionof the soils,
itativelyconsistentwith the observedoptical depthsderived discriminating
from advancedveryhighresolutionradiometer(AVHRR) sat- each with a correspondingsize distribution.Similarly,as by
Genthon [1992a, b], the sourcefunctionswere tuned afterellite (Husarand Stowe,unpublished
data, 1995).
Unfortunately, the AVHRR data give no information on wards to obtain the best agreementwith observationaldata.
dust aerosolin cloudedregions,and in this respectthey are a Some "shifts"of our sourcescomparedto thoseof Tegenand
lower limit to dust amounts. However, the calculated decrease Fung [1994] seemto be present,for example,our North Afriof Saharan dust concentrations
from the African
coast to
can sourceis strongerin the Sahel region,whereasTegenand
South America of about a factor of 10, is somewhat stronger Fung's[1994] sourceis strongestmore northerly.Our calcuhowever,agreeremarkablywell with
than the observed.In addition,no enhancedopticalthickness lated dustconcentrations,
is observedby satellite over South America and Australia, thoseof TegenandFung [1994],whichmaybe partly due to the
indicatingthat our model may be overestimatingthe dust a priori prescriptionof particlesizedistributionin our model
sourcesin these continents. On the other hand, measured dust
concentrationsat the South Pacific Islands (see Figure 5)
wouldbe underestimatedby an order of magnitudeby ignoring
these southernhemisphericsources.
Comparisonwith other models. Mineral aerosolhas been
modeledby three other groups,in varyingdegreesof detail.
Joussaume
[1990] used a general circulationmodel (GCM),
sourceregionsderivedfrom FAO statistics,and a sourcefunction basedon surfacedrag, to simulatethe cycleof mineral
aerosol.As only 1 aerosol size was considered,resultswere
presentedin arbitraryunits,and cannotbe compareddirectly
to our results,althoughgenerallymassconcentrationpatterns
are similarto our results.Genthon[1992a,b] simulateddustin
the GoddardInstitute for SpaceStudies(GISS) GCM using
and the a posterJorifit to observations
by Tegenand Fung
[1994].Recently,TegenandFung [1995]includedalsosources
inducedby land surfacemodification.Their calculationsindicate that the anthropogeniccontributionto the total global
mineral aerosol source may amount to 30-50%. In fact,
throughthe useof surfaceobservations
in our emissionfunction, thisanthropogenic
influenceis partlyaccountedfor (e.g.,
in the Sahel region), althoughelsewhere[e.g., India] an anthropogenicdust sourcemay be missing.
3.2. HeterogeneousReaction Rates on Mineral Aerosol
The mineral aerosoldispersedin the atmosphereprovides
reaction surfacesfor a variety of chemical/physical
processes,
asdiscussed
previously.
Onewayto illustratethe importanceof
22,878
DENTENER
I
60N
ET AL.' MINERAL
AEROSOL ON A REACTIVE
ar?nual axerage r,emoval rate on dust
I
i
i
0
SOE
_
SURFACE
I
......
:..
SON -
0.0z
30S
-
180W
150W
120W
,
,
90W
60W
SOW
60E
90E
120E
1 50E
1 80E
an,nual av,eracjec,olumn•tust surface
60N -
. ...............
•,•:..
......
SON-
•i::!
.::::.
..........
....
60S-
180W
:......
150W
120W
90W
60W
SOW
0
30E
60E
90E
120E
150F
180F
Figure8. Annualaverage(a) reactivity[xl03 s-1] on mineralaerosolcalculated
from aerosolsurface
Isolinesare 0.01, 0.03 0.07, 0.10,
distribution
for •/ = 0.1 (b) aerosolsurface
column[m ......
-2 • m earth
-2 surface]'
0.30, 0.70, and 1.00.
these surfaces is to calculate
removal
rates. Calculated
annual
averagedreactionratesusinga reactionprobabilityof •/= 0.1
is presentedin Figure 8a. Thesecalculatedcoefficientsmay be
comparedto the pseudo-first-orderrate coefficientscalculated
by Dentenet and Crutzen [1993] for the removal of N205 on
sulfate aerosol.
Calculated
removal
rates near the dust source
regionsappear to be of the sameorder of magnitudeas those
on sulfateaerosolin industrialregions(10-3 --10-4
able for reactionsmay amountup to 30% of the Earth's surface and in large parts of Africa and Asia this ratio is approximately 10%. As noted earlier this ratio may be considerably
(upto a factor of 10) higherif, for example,the aerosoldistributionby Shettle[1984]would be appliedin the sourceregions.
The resultingeffect on atmosphericchemistryof the heterogeneousremovalratesis stronglydependenton the gasphase
lifetime of specificcomponents.For examplethe OH radical,
with a typical lifetime of seconds,will be hardly directly af-
These high removal rates may also be interpreted in terms of
removalreactions
withratesof 10-3
aerosolsurfaceper squaremeter of underlyingEarth surface fectedby heterogeneous
[Figure 8b]. Above the sourceareasthe aerosolsurfaceavail- --10-4 s-1, althoughthere maybeindirecteffectsthrough
DENTENER
ET AL.: MINERAL
AEROSOL
ON A REACTIVE
SURFACE
22,879
60N
30N
E¸
30S
60S
t80W
150W
120W
i
90W
i
60W
i
30W
0
30E
90E
120E
150E
180E
60E
90œ
120E
150E
180E
.• 0.20
•0.20
30N
60E
i
-
EQ-
3OS -
60S
180W
150W
120W
90W
60W
30W
0
30E
Figure 9. Ratio of sulfatepresenton mineraldustto total sulfate(a) annualaverageand (b) FebruaryMarch-April. Isolinesare 0.0 to 1.0 in stepsof 0.1.
changesof other trace gasconcentrations.On the other hand,
ozonewith a typicalgasphaselifetime of 10 days,would be
stronglyinfluencedby the sameremoval rates. Note that the
actual •, valuesapplied in our model are given in section2.3.
3.3.
Influence of Dust on Sulfate, Nitrate,
and Ozone
Concentrations
Sulfate. Sulfateformationon mineral aerosolis presented
in Figures9a and Figure 9b. Theseshowthe annualaveraged
and February/March/April(FMA) averagedratios of sulfate
presenton mineral dustto total sulfateat the Earth's surface.
As explainedin section2, other sulfate-producing
reactionsin
our model (competingwith dust) are gas phaseoxidationby
OH and oxidationby H202 and 0 3 in clouddroplets[following
Langnerand Rodhe, 1992].
In the vicinity of the dust sourceregionswe calculatevery
largefractionsof sulfateassociated
with mineralaerosol,ranging from 50 to 70%. The regionswhereover 10% of the sulfate
is associated with the mineral aerosol extends from east Asia to
the central Pacific Ocean, and also to the southern Indian
Ocean.Further, it spansfrom centralAfrica to the northeastern parts of South America, and coversvast regionsof the
westernUnited States,southernSouth America, and Australia.
During the months of FMA (the high dust periods in the
northernhemisphere)the fractionof sulfateon mineral aerosol increases and extends farther
in the downwind
directions.
22,880
DENTENER
I
'
,
ET AL.: MINERAL
AEROSOL
ON A REACTIVE
SURFACE
an,nual
av?rage
in,
crease
,S04
du
3ON
i i:'
".,.
30S
180W
150W
130W
g0W
60W
30W
0
30E
60•
g0E
130E
150E
180E
Figure 10. Annual averageincreaseof total sulfatedue to reactionson mineral dust. Isolinesare 0.90, 1.00,
1.10, 1.25, 1.50, 1.75, and 2.0.
This is particularlyevidentin easternAsia, westernAfrica, and
in the tropical regions of the Indian Ocean. However, the
ratios of sulfate on mineral aerosol are only slightly higher
during the high dust season,indicatingthat the neutralization
reaction of SO2 is not limited by the dust Ca content.In fact,
especiallyin the boundarylayer, the reaction of SO2 with dust
is calculatedto be so fast that, compared to the reference
simulation without mineral aerosol surfaces,a significantly
larger fraction of SO2 reacts to sulfate instead of being removed by dry deposition.In Asia, where regionsof high SO2
emissionsare exposedto elevated dust concentrations,a significant increaseof the sulfate burden by up to a factor 2 is
calculated,with almost all the sulfate being present on the
surfaceof dust particles(Figure 10). This has important implications for the role of increasingsulfate in the radiative
forcing of climate. Having a larger fraction of the sulfate associatedwith the larger particlesmay diminishthe local cooling effectof the (accumulationrange) sulfateaerosol[Jonaset
al., 1995], whereas the sulfate present on the mineral dust
particleswill hardly change the radiative propertiesof those
dust particles.Present estimatesof sulfate coolingwhich ignore mineral aerosolsmay stronglyoverestimatethe forcing
(especiallyin Asia duringthe high dustseason).On the other
hand, the sulfateformed on the mineral aerosol(or attached
by coagulation)is likely to increasethe abilityof dustparticles
to act as CCN, and to enhancehaze, fog and cloud formation
[Parungoet al., 1995].This issuerequiresa more detailedstudy.
culations,and mostobservations
do not providerepresentative
climatologies.
Our resultsare confirmedby Mamane and Gottlieb [1989]
who showedthat a significantfraction of mineral aerosol is
coatedby sulfuricacidafter exposureto SO2.Earlier measurementsof Mamaneet al. [1980]in Israel haveshownthat during
dustdays63% of all sulfateis presentasmixedsulfates(desert
dust coatedwith sulfates).On clear days,this fraction is reduced to 20%. In this region we calculate ratios of around
50%. In Israel,Levin et al. [1990;Z. Levin et al., The effectsof
desert particles coated with sulfate on rain formation in the
easternMediterranean, submittedto the Journal of Applied
Meteorology,1995] find a fraction of the dust particlesto be
coated with sulfate. Interestingly,they observea rather con-
stantsulfursurfacedensity(in g /.cm-2),indicating
that the
mechanismof sulfateproductionis dependingstronglyon the
surface area. They interpret their measurementsmainly in
terms of "cloud processing,"but do not excludeother pathways,suchas reactionson dry or wetted aerosol.
Factor analysisby Winchesterand Wang [1989] on aerosol
from Asia and the Pacificprovidesstrongevidencethat almost
all aerosolsulfuris associatedwith soil elements.Also theyfind
that increasingsulfateamountsare presenton mineralaerosol
duringlongrangetransport.Zhanget al. [1993]performedsize
resolvedfactor analysisin five Chinesecitiesand theyfind both
fine (<2 /•m) and coarseaerosol to be enriched in sulfur,
althoughthe relative fractions are not presented.More reThe fraction of sulfate associated with dust in marine recently, single particle analysison dust collected at Qingdao,
gionsmay be overestimated,as SO2 reactionson seasalt aero- located500 km downwindof Beijing,showedthat 50 to 80% of
sol are likely to be more important [Lmia and Sievering;1991]. the large(d > 2/•m) particlesare coatedwith sulfate[Parungo
Further, cloudprocessing
of marine aerosolcangivea complex et al., 1995].
internalmixtureof si!icate.%
su!fate•and seasalt [Anreae eta!.,
.qi7f
• reqc•lved
measurementsin springat Yorita, Kagoshima,
1986];in sucha mixture sulfatecannotbe attributedsolelyto and Mt. Shibi in Japan [Horai et al., 1993],showfractionsof 5
mineral aerosol.Unfortunately,not many measurementshave to 40% of total sulfatebeing presenton particleslarger than
been performed in the regionsof interest to validate our cal- r = 0.5 /•m (cf, Figure 3). Most of this sulfate cannot be
DENTENER
ET AL.: MINERAL
AEROSOL
attributedto seasalt.It shouldbe noted (seesectionon effects
on the globalphotochemicaloxidantcycle) that in their work
the presenceof nitrate in the coarseaerosolfraction is even
more clearlyobserved.Gao et al. [1991]measureda substantial
fraction (20-30%) of sulfate to be present in coarsemode
aerosolin Japan.Hirai et al. [1991] proposed,basedon measuredrainwaterpH and Ca contentthat 75% and 90% of the
Kosa measured in Korea and Japan may be neutralized by
acids.Sizeresolvedmeasurementsin JapanduringKosa events
by Nishikawaet al. [1991a,b] showthat approximatelyhalf of
the total aerosol sulfate was present in the coarse aerosol.
Okada et al. [1987, 1990] found, usingX ray spectroscopy,
that
mostmineral particlesare coatedwith water solublematerial.
Indeed
we calculate
sulfate
mineral
aerosol
to total
sulfate
fractionsof 20-40% in these regions.
MeasurementsbyDaviset al. [1984]in 20 citiesin the United
Statesshow an averagecoarsemode sulfate fraction of 17%.
Elemental compositionshowsthat most of the coarsefraction
consistsof mineral aerosol.Wolff [1984] measureda smaller
coarse mode sulfate fraction of on average 7% for regions
more located toward the east. These measurementsroughly
agree with our calculations:howeverthey do not confirm our
calculatedhigh dust sulfate fractionsin the southwest.
Our resultsare contradictedby the observationsof Savoie
and Prospero[1982] who report sulfate mass in the tropical
Atlantic at Sal and Barbados,which are regionsof high dust
concentrations,to be mainly presentin the submicronaerosol
fraction.A possibleexplanationis that this accumulationmode
sulfate has been formed prior to encounterwith dust loaded
air. Wanget al. [1990] measuredsulfatein Beijing to be principily enrichedin the fine particle mode [0.5-1/•m] and not in
the dust coarsemode [4-8/•m]. Mukai et al. [1990] observed
that mostsulfateon Oki Islandnear Japanis presentassulfuric
acid and ammonium bisulfate, although again no quantitative
resultsare given.
ON A REACTIVE
SURFACE
22,881
nitrate) are presentedin Figures 11a and Figure 11b, respectively. The calculated ratio of dust nitrate to total nitrate is
calculatedto be almost unity in northern Africa, the Middle
East and Asia, indicatingthat the bufferingcapacityby mineral
dust is sufficientto neutralize the strongacidity by HNO3. The
regionswhere at least 40% of the total nitrate is found on the
mineral aerosolcoversvast regionsof the northern and southern Hemispheres.During the monthsof FMA the region covers almost all of Asia and extendsthroughoutthe the central
and northernregionsof the PacificOcean basin,and the tropical and subtropicalAtlantic and Indian Oceans.Only west and
central Europe, the eastern portions of North and Central
America, and the high latitude (>60 ø) zonesare predictedto
have relatively small fractions of HNO3 associatedwith the
mineral aerosol. In addition, regions of strong gradients of
nitrate on dustare predictedin the subtropicalAtlantic Ocean,
and the low- and midlatitude regionsof the western Pacific.
A strong correlation of high total nitrate with high dust
concentrations
on the Atlantic
and
Pacific
ocean
has been
noted by severalauthors[e.g.,Savoieet al., 1989;Prosperoand
Savoie,1989;Prosperoet al., 1995].This correlationis primarily
due to the same meteorologicalconditionsunder which both
dust and photochemicallyproduced nitrate are transported
from the continental source regions. It is unlikely, however,
that significantamountsof gas phase HNO3 can survivelong
transport distances,due to the high dry depositionvelocityof
HNO3. Nitrate on the mineral dust surface can, however, be
transportedlong distancessince these particles have a lower
dry depositionvelocity (at least for aerosolssmaller than 10
Unfortunately, like for sulfate,there are relativelyfew measurementsof aerosol nitrate available in the regions under
consideration,and very few simultaneousmeasurementsof gas
phaseHNO3 and aerosolnitrate. Analysisof aerosoland gas-
eousNOyandNOxmeasurements
at ChejuIsland,S. Korea,
usingaerosoldynamicsand equilibriummodelsindicatesthat
tively large fraction of aerosolsulfatebeing present in coarse 20 to 60% of the nitrate is expectedto be in the aerosolphase
aerosol,at least in the vicinityof dustsourceregions.It should [Carmichaelet al., 1995].Thesevaluesare consistentwith those
be noted that we did not focus on sulfate measurements
in
presentedin Figure 11.
As shown previouslyin Figure 2a, Asian aerosol showsa
regionsof knownlow dustconcentrations.In the latter regions,
sulfate is likely to follow a typical accumulationrange size strong correlation between aerosol nitrate and calcium. The
distribution.
few existingsize resolvedmeasurementsin Korea and Japan
It hasbeen questionedwhether sulfatefound in coarseaero- [e.g.,Horai et al., 1993; Gao et al., 1991;Nishikawa,1991a,b]
sol is derivedfrom reactionsduringthe long-rangetransportof showsignificantamountsof nitrate presentin the coarsefracmineral aerosol,or originatingfrom the soil itself.Nishikawaet tion aerosol(cf, Figure 3). In the USA and Israel, Ganor and
al. [1991b] present sulfate weight fractions in soils of arid Pueschel[1988] found many coarsefraction mineral aerosols
regionsin China varying from less then 0.01% to 0.46% (S to be coatedwith nitrate. Similarly, Wolff [1984] measuredon
content0.003-0.15%). Clearlythisis insufficientto explainthe average 67% of aerosol nitrate to be present in the coarse
high coarsefraction sulfate concentrationsfound in mineral fraction aerosol.Parungo et al. [1995], found that at RH >
aerosol.Likewise, soil sulfate contentsof the Sahara are gen- 50%, more than 50% of the large particlesat Qingdao, China,
erally thought to be low [Lo•e-Pilot et al., 1986]. Still some were coated with nitrates, while less than 20% of the small
gypsiferous
soilsmay existin the Thar Desert, Iranian and and particlescontained nitrate.
The consequenceof HNO3 being presenton mineral dust,
Arabian deserts[Dregne,1968, as referencedby Savoieet al.,
1987]. It is not clear, however,whether thesesoil sulfatecon- insteadof in the gas phaseor in accumulationrange aerosol,
tents refer to the bulk mass, or to the fraction available for may be significant.For example,HNO3 reactingwith mineral
uplift of the soils,the latter obviouslybeing most relevantfor aerosolin or closeto the dust sourceregionsmay be removed
our calculations.Finally, anotherfeature, not well represented faster than gas phase HNO3 due to sedimentationof coarse
in our model, is the patchynature of high dustconcentrations. mode aerosol. On the other hand, HNO 3 associatedwith the
This may lead to coexistingsubgridregionsof high dust and fractionof mineral aerosolsubjectto long rangetransport,may
high SO2 concentrations,and overestimatethe reaction be- substantiallyenhance the transport distance of nitrate, and
tween them.
thus provide a means for HNO 3 to be transportedto remote
Nitrate. The annually averagedand FMA averaged ratio atmosphericregions,where it can participatein photochemisof NO•. However, we did not atof nitrate on dust to total nitrate (=gas phaseHNO3 + dust try after photodissociation
All in all, most observations confirm our results of a rela-
22,882
DENTENER
i
60N
i
ET AL.: MINERAL
AEROSOL ON A REACTIVE
SURFACE
aqnual av,erage rgtio HNQ5 on dust to t9tal HN95
-
50N
EQ
3OS
•
1 80W
1 50W
120W
,
90W
,
................
60W
30W
0
30E
90E
60E
120E
150E
180E
YMA,ratio H•103 on, dust to, total H,N03
::::::::::::::::::::::::::::::::::::::::
60N
..
30 N
0.40
3OS
o
.20
60S
180W 15'OW120W 90W 60W 30W
0
30E
60E
90E 120E 150E 180E
Figure 11. Ratio of nitrate presenton mineral dust to total nitrate (a) annual averageand (b) FebruaryMarch-April. Isolinesare 0.20, 0.40, 0.60, 0.80, 0.90, 1.00, and 1.10.
surfacemay also take place, as discussedpreviously.In this
sectionwe evaluate the effectsof the reactionsof N:Os, HO:,
and O_•on dust particleson ozone levels.
interaction with nitric acid as mineral aerosol. Thus, in the
Boundarylayer ozoneis reducedby about 10% by including
marine boundarylayer the high fraction of nitrate associated thesereactionsin the dust sourceregionsduring the high dust
with mineral aerosolpresentedin Figures 11a and 11b may in season,and up to 8% yearly averaged(Figure 12a and 12b).
reality'be associatedwith sea salt aerosol.
Outside the high dust regionsozone decreaseswere found to
Effectson the global photochemicaloxidant cycle. Mineral be very small. The HO: radical concentrationsare calculated
aerosolsmay affectthe photochemicaloxidantcycleby provid- to decreaseby about 10% in the dust sourceregions(Figure
ing additional reactionpathwaysfor speciessuchas N20 s and 13). However, the contributionof this reducedHO: concenHO2, whichin turn can influenceozoneproduction/destruction trationsto O_•decreaseis negligible,as the regionsof highest
rates. In addition, direct destruction of ozone on the dust dust concentrationsgenerally have low NOx concentrations,
tempt to assessthe significanceof this processin this paper. It
shouldbe noted that over the oceans(at least in the marine
boundarylayer) seasalt aerosoloffersa similarpossibilityfor
DENTENER
,
60N
,
ET AL.' MINERAL
AEROSOL ON A REACTIVE
SURFACE
22,883
,annualgverage,reductignof 0•3 (all re,actionsI
-
30N -
E¸-
30S
-
60S
-
_
180W
150W
t20W
,
,
90W
60W
-
30N
-
0
$0E
60E
90E
120E
150E
180E
F•A ratio,of ozo9e due •o dust ,reaction,
s surfa9e
B
60N
SOW
.............
::;
:.:::
:•::
'";;
:"'::•';•::;•
%:
....
.
:::
:::::::
:::
::
ß
•::.;:::
•:.::::::::::::::::::::::
::.:.q:
::,.u.•.::.
.....•
• ::•::::: :.:•:::
:::.:.:...:•.::::.:::::
::
3OS -
:•.
;f::::.:::
::•::•:: ....
•::
.4::•:::
60S
180W
150W
120W
90W
60W
30W
0
30E
60E
90E
120E
150E
180E
Figure 12. Calculated ratio of ozone indudin• interactionsof as, NO•, and HO• with dust to a dust-free
situation(a) annual averaBe
, (b) Februa•-March-April, (c) the calculateddecreaseof as in Februa•-
March-April,considerinB
an upperlimit of ?(O•) - 2 x 1• •. Isolinesare •.9•, •.92, •.94, •.96, and•.98.
and photochemicalproduction of 03 by the reaction of NO
with HO 2 is slow.
Most of the calculateddecreaseof 03 (2-6%) is causedby
the assumeddirect 03 destructionreactionon aerosol.As has
been argued before, the destructionrates of 03 on mineral
aerosolare extremelyuncertain;lower ,/values yield negligible
destructionand higher valueshavingsome importance.Using
Destruction of N205 on mineral aerosol could by itself be
effectivein removingNOx from the atmosphere.However, in
our calculations[followingDentenerand Crutzen, 1993], we
used, in addition to mineral aerosol, an accumulationrange
"background" sulfate concentration, on which N205 is also
removed.As a consequenceof the SO2 reactionson dust, this
"background"sulfate decreased,and the role of sulfate in
the estimated
upperlimit of ,/(03) of 2 x 10-4, whichis 4 removing N205 was taken over by the dust. The removal of
timesashighasour "bestestimate"for ,/(03), ozonedecreases N205 on dust therefore had only a small "additional" effect
would amount to about 20% in vast regions of North Africa, compared to our reference simulation, which ignored the effects of dust.
the Arabian Peninsula,and Middle Asia (Figure 12c).
22,884
DENTENER ET AL.: MINERAL AEROSOL ON A REACTIVE SURFACE
,
,
FM/•reductionof 0,5 (gam,ma-2Ei-4 )
120W
90W
,
C
60N
-
5ON -
0.98
180W
150W
60W
50W
0
50E
60E
90E
120E
150E
180E
Figure 12. (Continued)
ontheorderof 3-10%per10/•g m-3 dust.On
Our resultscanbe comparedwith the boxmodelsimulations 03 decreases
of photochemistry
duringthe outbreakof Kosaduststormsby synopticalscales,however,high dust and 0 3 concentrations
Zhanget al. [1994].They estimatedozonedecreases
of about are correlated.Although the short-termanticorrelationof 03
conditions,our
10% due to reactionson mineral dust.They did, however,not maybe due in part to specificmeteorological
consider the direct ozone destruction on aerosol, so that the
calculated03 decreasesof a few percent for dust concentra-
effectsof N205 and odd hydrogenreactionson dust are estimated to be somewhatlarger than in this study.
Measurementsin Happo, Japan,presentedby Zhanget al.
tionsof 10-100/xgm
-3 is consistent
withtheseobservations.
[1994],showa short-termanticorrelation
of O3withdust,with
Regionalresults. So far in this paper we have restricted
our discussion
to the globalimpactof mineral aerosols.However, the effectsdiscussedabovecan be intensifiedon regional
FMA de,crease gf H02 ,surface,
60N
-
50N
-
EQ-
30S
-
60S
-
i
i
i
i
i
i
i
180W
150W
120W
90W
60W
50W
0
i
50E
i
60E
i
90E
i
i
i
120E
150E
180E
Figure 13. Ratioof HO 2 concentrations
withdustreactions
compared
to a dustfreesituationin FebruaryMarch-April. Isolinesare 0.80,0.82,0.84,0.86,0.88,0.90,0.92,0.94,0.96,and0.98.
DENTENER
ET AL.' MINERAL
AEROSOL
ON A REACTIVE
SURFACE
22,885
56
ß 47
50.
100.
200.
MAX
400.
'-
• 38
J
(Unit:ppb)
99
1 08
117
1 26
I 35
1 44
1 53
162
Long i rude
0.8
8
0
7
0
6
0
5
0
:•:.•
......
-•
4
0
•
ß
3
0
2
0
1
0
km
-""'"•"'•'
'•g)•-::.•::
:-?"•""
'"•'•
' :'::
-:•
.... :-'•'"•"
"'•'½'•-"•'-'4•'
:-'-a:'•'
::•': .....
•,i•:.:::::...::
:i!:
-:
....
:. ::'::
'"•::• ....
'"'"'-'-"•--'
: ......
'•'"'"'""'"'•J•$...
?'•""•4•!:.
'%!i':-::'?•?':•:•:':'":":•
!:i:.,.-6:•:?•½•(::::.:::
":"":'"':"2•;5:•.:•:!:?:"
. •;!i:.•:iil
i
_ ---.•.::.--::.•:::•-'-;
•..'.:'•57:•,:•...:•
.........
...:..•:•:.::...
..•.•:::-.:•.•.-.-::-.
.:--•.•
-:,,.
--...-•..
99
....
:i'
,
:i',:•,
' I I
105
112
..-.'
:.:.:.:
_
"':'"":•-.Y•......
Y:......
118
124
I
131
I .."::::"1'
137
143
149
15•
162
99
Long i rude
latitude-4•
105
112
118
124
131
137
143
149
156
162
Long i rude
latitude=35
Figure 14. Calculated dust distribution on May 6, 1987, using the regional scale STEM-1I model. A series
of dust stormsoccurredduring the first 2 weeks in May.
scalesduring high dust periods. For example, consider the
situation in east Asia, where dust stormsare common during
the springtime. Calculated mineral dust concentrationsfor a
typicalduststormduringMay 1987are presentedin Figure 14.
These resultsare calculatedusingthe STEM-II regional scale
transport/chemistry
model [Carmichaelet al., 1991], with a 1ø x
1øresolutiondrivenby ECMWF data. As shown,the calculated
Similar effects can be anticipated off the coast of West Africa, and other locations where dust storms are a common
occurrence.This will be the subjectof a future paper.
4.
Summary and Conclusions
In this paper we explored whether mineral aerosol can imconcentrations
of mineralaerosolexceed400 •g/m3 at loca- pact troposphericchemistry. We explored this issue using a
tionsthousandsof kilometersawayfrom the sourceareas,and global, three-dimensional model of the troposphere, which
the effectsof the duststormare felt throughoutthe middle and coupled mineral dust processeswith the photochemical,nitrolower troposphere.During this particular event, concentra- gen, and sulfur cycles.There is a large body of observational
tionsin excessof 75 •g/m3 persisted
for over5 days.These data that supportsthe idea that mineral aerosolsare an imconcentrationsare 10 to 20 times higher than the monthly portant reactive surface.However, very little is known regardaveragedvaluesshownin Figure 4b. These large surfaceareas ing the mechanismsby which mineral aerosolsimpact these
can exert a significantimpact on the regional biogeochemical cycles.
Mechanisms
for the interactionof SO,,,NOy, and 0 3
and photochemicaloxidant cyclesin eastern Asia. For exam- with mineral surfaceshave been postulated,and used to calple, ozone levels in the presenceof the elevated dust levels culate heterogeneousremoval rates. These calculationsremain
decreasedby 20 to 50% when the dust-chemicalinteractions highly uncertain, mainly due to the fact that these surfaces
were included in the calculation. In addition, the effects of the have not been widely studied from a reactive surface standdust perturbations on ozone remained for 8 days. Figure 15 point, and basicinformation on sorptionand reaction rates are
illustratesthese effects.Shown are calculatedboundary layer lacking. Especiallythe most uncertain processesin our study,
concentrations
of dust and ozone concentrations
with and
the direct uptake of ozone on dust particles and the mechawithout the dust-chemistryinteractionsfor a locationin central nism and rate of uptake of SO2 on dust, deserve priority in
Japan.When the dust-chemistryinteractionsare included,not laboratory and field experiments.Removal rates are quite deonly are the ozone levelsreduced significantly,but the corre- pendant on the deliquenscenceproperties of dust particles.
lation between ozone and dust is significantlyshifted from These properties have been measured by Hiinel [1976] for
positiveto negative,in accordwith thosereported by Zhang et African dustaerosol,but it is not clear if they can be applied to
al. [1994].
other regions.
22,886
DENTENER
ET AL.' MINERAL
AEROSOL ON A REACTIVE
SURFACE
140
140
03,
with
dus
120
120
lOO
IO0
_;_
.....
o,,;tno
80 •
,'",
N
o
,,'
',
---',, 'f
60
I ',/. /
40-
60 •
,,
"-",_:'"-"
"-",.>C/"
20-
ob
40
20
'o
987
Figure 15. Calculated boundary layer concentrationsof dust and ozone with and without dust-chemistry
interactionsfor a site in central Japan [Zhanget al., 1995].
In addition
to the reduction
of uncertainties
in kinetical
parameters,this work would improveby a more detailed meteorologicaltreatmentof dustuplift, and consequenttransport
in the atmosphere,for example,followingthe modelsof Tegen
and Fung [1995], and a better understandingof the dustsizedistributions[e.g.,Dulac et al., 1992]. There are someuncertainties associated with the sulfate, nitrate, and calcium car-
bonate contentassociatedwith the dustparticlesavailablefor
uplift in the sourceareas.Additional measurements
would be
months of FMA the region covers almost all of Asia and
extendsthroughoutthe central and northern regionsof the
PacificOcean basin, and the tropical and subtropicalAtlantic
and Indian Oceans.Only the regionsof western and central
Europe, the easternpartsof North and Central America, and
the highlatitude(>60 ø) zonesare predictedto haverelatively
small portionsof HNO 3 associatedwith mineral aerosol.The
consequences
of HNO3 uptake
onmineral
aerosol
forpho-
tooxidant chemistryare not clear. One possibilityis that increasedtransportdistancesof nitrate on mineral aerosolmay
Our resultsillustrate that mineral aerosolcan have a signif- affect the photochemistryin remote areas. In this work we
icant impacton the chemistryof the troposphere.In the caseof assumedthat uptake of HNO 3 on mineral aerosol did not
sulfate,a significantfraction of sulfateis predictedto be asso- hinderuptakeof SO2,other than depletingthe CaCO3 content
ciated with mineral aerosol.The regionswhere over 10% of of the aerosol.It would be valuable to verify this.
the sulfate is associated with mineral aerosol extend from east
Interactionsof N205, 03, and HO 2 radicalswith dustwere
Asia to the central Pacific Ocean, and out into the southern also found to affect the photochemicaloxidant cycle, with
Indian Ocean, spansfrom central Africa to the northeastern ozone concentrationsdecreasingby up to 10% in and nearby
portions of South America, and covers vast regions of the the dust source areas. Comparisonof these resultswith the
western United States, southern South America, and Australia.
limited available measurementsindicates that the proposed
The neutralization reaction of SO2 does not appear to be reactionscan indeedtake place, althoughthe lack of measurelimitedby the dustCa2+ content.In factthe reactionof SO2 mentspreventsa rigorousvalidation.
The direct reaction of ozone on mineral aerosol warrants
with dust was calculatedto be so fast that, comparedto the
referencesimulationwithout mineral aerosolsurfaces,a sig- further study,and could be important at accommodationconificantlylarger fraction of SO2 was found to react to sulfate efficients
exceeding
5 x 10-5. Theseeffectscanbe intensified
instead of being removed by dry deposition.These results on regional scalesduring high dust periods, where surface
suggestthat the previousassumptionthat in East Asia sulfate areas of the mineral aerosol can be an order of magnitude
aerosolfollowsan accumulationrange aerosolsizedistribution higher during dust storm comparedto the monthly averaged
is likely to overestimatethe sulfate aerosol climate-cooling values calculated by our global model. In addition, tropoeffect. On the other hand, these processesmay enhancethe sphericreactionson dustcan be expectedto play a larger role
CCN activity of the mineral aerosol. The nucleation of cloud duringclimaticperiodsof higher dustloadings.For example,
dropletson mineral particlesand subsequentcloud chemistry during the last glacial maximum (---18,000years ago), dust
and oxidation of SO2 has been proposed as an important levelswere up to an order of magnitudehigherthan thoseused
mechanism
by Z. Levin et al. (1995).This processhasnot been in this paper [cf, Petit et al., 1981].
accountedfor in our model,but certainlydeservesmore attenMineral aerosolsthus appear to provide an important reaction.
tive surface. Clearly, more research is necessaryto further
An even larger fraction of gas phase nitric acid may be quantify the role of these aerosolsin the chemistryof the
neutralized by nfineral aerosol, and there is less uncertainty troposphere.
about this processcomparedto the conversionof SO2 to suifate on dust particles.The parts of the globe where at least
useful.
40% of the total nitrate
is found on mineral
aerosol cover vast
regionsof the northern and southernhemispheres.During the
Acknowledgments. F.D. likes to thank the University of Iowa for
support during his visit to Iowa and the Max-Planck-Instititut fdr
DENTENER
ET AL.: MINERAL
AEROSOL
Chemie, Mainz, Germany, where this work was initiated. We appreciate the discussionsand comments of Yves Balkanski, Frank Raes,
Lothar Sch0tz,David Sedlak,Peter Zimmermann, and two anonymous
reviewers.The authors thank M. Wefers for making his dust model
available to us. This researchwas supportedin part by the German
SFB, NASA (Grant NAGW-2428), and the European Communities
projectSINDICATE. G. C. and Y. Z. wishto extendspecialrecognition Ching-Yi Chang and Hioshi Hayami for their assistancein the
preparationof this manuscript.
ON A REACTIVE
SURFACE
22,887
Mediterranean sea usingmeteosatdata, J. Geophys.Res.,97, 24892506, 1992.
Feichter, J., and P. J. Crutzen, Parameterisation of deep cumulus
convectionina globaltrace transportmodel and its evaluationwith
222Radon,TellusB, 42, 100-117, 1990.
Fendel, W., D. Matter, H. Burtscher, and A. Schmidt-Ott, Interaction
between carbon or iron aerosol particles and ozone, Atmos. Environ., 29, 967-973, 1995.
Fenter, F. F., F. Caloz, and M. J. Rossi, Heterogeneous kinetics of
N20 s uptake on salt,with a systematicstudyof the role of surface
presentation(for N205 andHNO3), J. Phys.Chem.,100.,1008-1019,
References
Aldaz, L., Flux measurementsof atmosphericozone over land and
water, J. Geophys.Res., 74, 6943-6946, 1969.
d'Almeida, G. A., On the variability of desert aerosolradiative characteristics,J. Geophys.Res.,92, 3017-3026, 1987.
Andreae, M. O., Climatic effects of changing atmospheric aerosol
levels,in WorldSurveyof Climatology:
FutureClimatesof the World,
vol. 16, edited by A. Henderson-Sellers,Elsevier, New York, 1995.
Andreae, M. O., Raising dust in the greenhouse,Nature, 389-390,
1996.
Andreae, M. O., R. J. Charlson, F. Bruynseels,H. Storms, R. van
Grieken, and W. Maenhout, Internal mixture of seasalt, silicates,
and excesssulfate in marine aerosols,Science,232, 1620-1623, 1986.
Beilke, S., and G. Gravenhorst,HeterogeneousSO2 oxidation in the
droplet phase,Atmos.Environ.,12, 171-177, 1978.
Benkovitz, C. M., M. T. Scholtz, J. Pacyna, L. Tarrason, J. Dignon,
E. C. Voldner, P. A. Spiro,J. A. Logan, and T. E. Graedel, Global
griddedinventoriesof anthropogenicemissionsof sulfur and nitrogen,J. Geophys.Res.,in press,1996.
Carmichael, G. R., L. K. Peters, and R. D. Saylor, The STEM-II
RegionalScaleAcid Depositionand PhotochemicalOxidantModel,
I, An overviewof model developmentand applications,Atmos.Environ., 25, 2077-2090, 1991.
1996.
Fuchs,N. A., and A. G. Sutugin,Highly DispersedAerosols,105 pp.,
Butterworth-Heinemann,
Newton, Mass., 1970.
Fujita, S., and A. Takahashi,Acidic Depositionin East Asia, 35 pp.,
Rep. T93091, Central Inst. for Electric Power Ind., Tokyo, Japan
1994.
Galbally, I. E., and C. R. Roy, Destructionof ozone at the earth's
surface, Q. J. R. Meteorol. Soc., 106, 559-620, 1980.
Ganor, E., and R. F. Pueschel,Compositionof individualnitrate containingparticlesin non-urbanatmospheresof Colorado,USA and
Tel Aviv, Israel, Water Air Soil Pollut., 42, 169-181, 1988.
Gao, Y. C., D. Zhang, and T. Okita, Size distributionof atmospheric
aerosolsand their solubleion speciesin Mei-Long area, Shanghai,
Proc. 2nd IUAPPA Reg. Conf Air Pollut. 1, 65-70, 1991.
Garland, J. A., Dry depositionof SO2 and other gases,Proc. Atmos.
Surf. ExchangeParticul. Gaseouspollut., 1974 Syrup.,U.S. Dep. of
Commerce, 1976.
Genthon, C., Simulationsof the long range transport of desert dust
and sea-saltin a generalcirculationmodel in Precipitation,ScavengingandAtmospheric
SurfaceExchange,editedby S. E. Schwartzand
W. G. N. Slinn, HemispherePubl. Washington,1992a.
Genthon, C., Simulations of desert dust and sea-salt aerosolsin Antarcticawith a generalcirculationmodel of the atmosphere,Tellus,
44B, 371-389, 1992b.
Carmichael,G., Y. Zhang, L. Chen, M. Hong, and H. Ueda. Seasonal
variation of aerosol compositionat Cheju Island, Korea, Atmos. Gillette, D. A., G. J. Stensland,A. L. Williams, W. Barnard, D. Gatz,
P. C. Sinclair, and T. C. Johnson, Emissions of alkaline elements
Environ., in press, 1995.
Chameides,W. L., and A. Stelson,Aqueous-phasechemicalprocesses
calcium,magnesium,potassium,and sodiumfrom open sourcesin
in deliquescentsea-saltaerosols:A mechanismthat couplesthe
the contigeousUnited States, Global Biogeochem.Cycles,6, 437atmosphericcyclesof S and sea salt,J. Geophys.Res., 97, 20,565457, 1992.
20,580, 1992.
G•sten, H., G. Heinrich, E. M6nnich, D. Sprung,J. Weppncr, A. B.
Charlson, R. J., S. E. Schwartz,J. M. Hales, R. D. Cess,J. A. Coakley,
Ramadan, M. R. M. Ezz El-Din, D. M. Ahmed, and G. K. Y.
JR, J. E. Hansen, and D. J. Hofmann, Climate forcing by anthroHassan, On-line measurementsof ozone surface fluxes, II, Surface
pogenicaerosols,Science,255, 423-430, 1992.
level ozone fluxes onto the Sahara desert, Atmos. Environ., 30, 911Chiapello, I., G. Bergamatti, F. Dulac, L. Gomes, B. Chatenet, J.
918, 1995.
Pimenta, and E. S. Suares, An additional low layer tranport of
Hanson,P. J., and S. E. Lindberg,Dry depositionof reactivenitrogen
Sahelianand Saharandust over the North-Eastcrn tropical atlantic,
compounds:a reviewof leaf, canopyand non-foliar measurements,
Geophys.Res. Lett., 22, 3191-9194, 1995.
Atmos. Environ., 25A, 1615-1634, 1991.
Davis, B. L., L. R. Johnson,R. K. Stevens,W. J. Courtney, and D. W.
Hanson, D. R., and E. R. Lovejoy, The uptake of N205 on small
Safriet, The quartz content and elemental compositionof aerosols
sulfuricacid particles,Geophys.Res.Lett., 21, 2401-2404, 1994.
from selectedsitesof the EPA inhalableparticulatenetwork,Atmos. Hanson, D. R., J. B. Burkholder, C. J. Howard, and A. R. RavishanEnviron., 18, 771-782, 1984.
Dentener, F. J., and P. J. Crutzcn, Reaction of N20 s on tropospheric
aerosols:impacton the globaldistributionsof NOx, 03, and OH, J.
Geophys.Res.,98, 7149-7163, 1993.
Dentener, F. J., and P. J. Crutzen, A Three-dimensional Model of the
Global Ammonia Cycle,J. Atmos. Chem., 19, 331-369, 1994.
Dlugi, R., S. Jordan, and E. Lindemann, The Heterogeneousformation of sulfate aerosolsin the atmosphere,J. AerosolSci., 12, 185197, 1981.
Dregne, H. E., Appraisalof researchsurfacematerialsof desertenvironmentsin Desertsof the World,an Appraisalof Researchinto Their
Physicaland BiologicalEnvironments,edited by W. G. McGinnies et
al., pp. 287-377, Univ. of Ariz. Press,Phoenix,1968.
Duce, R. A., Source, distributions, and fluxes of mineral aerosols and
their relationshipto climate,in Dahlem Workshop
onAerosolFo•ving
of Climate, edited by R. J. Chaffson and J. Heintzenberg, John
Wiley, New York, 1995.
Duce, R. A., C. K. Unni, B. J. Ray, J. M. Prospero,and J. T. Merrill,
Long-range atmospherictransport of soil dust from Asia to the
tropicalNorth Pacific:Temporalvariability,Science,209, 1522-1524,
1980.
kara, Measurement of OH and HO: uptake coefficientson water
and sulfur acid surfaces,J. Phys.Chem., 96, 4979-4985, 1992.
Hfinel, G., The propertiesof atmosphericaerosolparticlesasfunctions
of relative humidity at thermodynamicequilibrium with the surroundingmoist air, Adv. Geophys.,19, 73-125, 1976.
Hirai, E., M. Miyazaki, T. Chohji, M. Lee, M. Kitamura, and K.
Yamaguchi,Effect of Kosa aerosolon inorganicion componentsin
rainwatercollectedfrom circum-pan-japan-sea
area,Proc.2nd IUAPPA Reg. Conf Air Pollut., 2, 27-34, 1991.
Horai, S., I. Minari, and Y. Migita, Aerosols Composition in Kagoshimain theAnnual Reportof theKagoshimaPrefecturalInstitute,
vol. 9, 1993.
Iwasaka,Y., I. Mori, K. Matsunaga,Y. Mural, M. Hayashi,T. Shibita,
andT. Ojio, Aircraft measurements
on KOSA overJapanIslands,in
Proceedings
of theInternationalConference
on RegionalEnvironment
and ClimateChangesin EasternAsia, edited by C. M. Li, pp. 49-53,
1993.
Jaeger,L., Monatskartedes Niederschlagsfor die ganze Erde, Ber.
Deutsch. Wetterdienstes,139, 1976.
Jaenicke, R., Landolt-BornsteinZahlenwerte und Funktionen aus Natur-
Duce, R. A., et al., The atmosphericinput of trace speciesto the world
ocean,Global Biogeochem.Cycles,5, 193-259, 1991.
Dulac, F., D. Tanrd, G. Bergametti,P. Buat-Mdnard,M. Desbois,and
wissenschaften
und Technik,4, Meteorologie,
B, Physikalische
und
ChemischeEigenschaften
der Luff, 391-457, Springer,New York,
D. Sutton, Assessmentof the African dust mass over the Western
Jonas, P., R. Charlson, and H. Rodhe, Aerosols in Climate Change
1988.
22,888
DENTENER
ET AL.: MINERAL
AEROSOL ON A REACTIVE
SURFACE
1994, edited by Houghton et al., CambridgeUniv. Press,New York,
aerosolas eolian carrier of anthropogenicmaterial, Sci. Total Envi-
1995.
ron., 107, 13-27 1991b.
Joussaume,S., Three dimensionalsimulationsof the atmosphericcycle Ohta, K., Aerosolchemicalcompositionof Kosa,in Kosa,editedby K.
of desertdustparticlesusinga generalcirculationmodel,J. Geophys.
Higuchi, Kokon-shoin,Tokyo, Japan 1991.
Res., 95, 1909-1941, 1990.
Okada, K., A. Kobayashi, Y. Iwasaka, H. Naruse, T. Tanaka, O.
Judeikes,H. S., and A. G. Wren, Laboratory measurementsof NO and
Nemoto, Featuresof individualAsian dust-stormparticlescollected
NO 2 depositionsonto soil and cement surfaces,Atmos.Environ., 12,
at Nagoya, Japan,J. Meteorol.Soc.Jpn., 65, 515-521, 1987.
2315-2319, 1979.
Okada, K., et al., X-ray spectrometryof individual asian dust storm
Judeikes,H. S., T. B. Stewart, and A. G. Wren, Laboratory studiesof
particlesover the JapaneseIslands and the North Pacific Ocean,
Atmos. Environ., 24A, 1369-1378, 1990.
heterogeneousreactions of SO2, Atmos. Environ., 12, 1633-1641,
1978.
Oort, A. H., Global atmosphericcirculation statistics,1958-1972,
Judeikes,H. S., S. Siegel, T. B. Stewart, H. R. Hedgpeth, and A. G.
NOAA Prof. Pap., 14, US Governmentprinting office,Washington
DC, 1983.
Wren, Laboratory studiesof heterogeneousoxidation of oxides of
nitrogenin Nitrogeneous
Air Pollutants,Chemicaland BiologicalImParrington, J. R., W. H. Zoller, N. K. Aras, Asian dust: Seasonal
plications, edited by D. Grosjean, pp. 83-109, Butterworthtransportto the Hawaiian Islands,Science,220, 195-197, 1983.
Heinemann, Newton, Mass., 1979.
Parungo,F., Y. Kim, C-J Zhu, J. Harris, R. Schnell,X-S Li, D-Z yang,
M-Y Zhou, Z. Chen, and K. Park, Asian dust storms and their
Junge,C. E., and P. E. Gustafson,On the distributionof seasaltover
effectson radiation and climate, STC Rep. 2906, Natl. Oceanic and
the United States and its removal by precipitation,Tellus,9, 164173, 1957.
Atmos. Admin. Air Resour. Lab, Silver Spring, Md., 1995.
Kang, K. H., and S. E. Sang,influenceof yellow•and on TSP in Seoul, Payrissat,M., and S. Beilke, Laboratorymeasurementsof the uptake
Proc. 2nd IUAPPA Reg. Conf.Air Pollut., 2, 1-6, 1991.
of sulphurdioxideby different European soils,Atmos.Environ.,9,
211-217, 1975.
Karyampudi,V. M., and T. N. Carlson,Analysisand numericalsimulations of the Saharan air layer and its effects on easterly wave Petit, J., M. Briat, and A. Royer, An ice-ageaerosolcontentfrom East
disturbances, • Atmos. e•; ,,½ •,•
•,
,•
Antartic ice core samplesin past wind strengths,Nature, 293, 391394, 1981.
Kiehl, J., and B. Briegleb, The relative roles of sulfate aerosolsand
Preining,O., Aerosol and climate-An overview,Atmos.Environ.,25A,
greenhousegasesin climate forcing,Science,260, 311-314, 1993.
2443-2444, 1991.
Langner, J., and H. Rodhe, A global three-dimensionalmodel of the
Press,W. H., S. A. Flannery, S. A. Teukolsky,and W. T. Vetterling,
troposphericsulfur cycle,J. Atmos. Chem., 13,225-13,265,1991.
Lelieveld, J., and P. J. Crutzen, Influences of cloud photochemical
NumericalRecipes,CambridgeUniv. Press,New York, 1986.
Prospero,J. M., Aeolian transportto the world oceanin, The Sea,vol.
processes
ontropospheric
ozone,
Nature,343,2•7-233,1990.
Levin, Z., C. Price, and E. Ganor, The contribution of sulfate and
7, The Oceanic Lithosphere,edited by C. Emiliani, pp. 801-974,
desert aerosols to the acidification of clouds and rain in Israel,
Wiley Intersci., New York, 1981.
Atmos. Environ., 24, 1143-1151, 1990.
Prospero,J. M., Mineral and seasalt aerosolconcentrationsin various
Li, S. M., K. G. Anlauf, and H. A. Wiebe, Heterogeneousnighttime
oceanregions,J. Geophys.Res.,84, 725-731, 1979.
productionand depositionof particlenitrate at a rural sitein North Prospero,J. M., and D. L. Savoie, Effect of continental sourceson
nitrate concentrationsover the Pacific Ocean, Nature, 339, 687-689,
America duringsummer1988,J. Geophys.Res.,98, 5139-5157, 1993.
1989.
Li, X., H. Mating, D. Savoie,K. Voss, and J. M. Prospero,Dominance
of mineral dust in aerosol light-scatteringin the North-Atlantic Prospero,J. M., D. L. Savoie,T. N. Carlson, and R. T. Nees, Monitrade winds, Nature, 380, 416-419, 1996.
toring Saharanaerosoltransportby meansof atmosphericturbidity
measurementsin SaharanDust: Mobilisation,Tranport,edited by C.
Loye-Pilot,M.D., M. Martin, andJ. MoreIll, Influenceof Saharandust
on the acidity and atmosphericinput to the Meditteranean,Nature,
Morales, pp. 171-191, John Wiley, New York, 1979.
321,427-428, 1986.
Prospero,J. M., R. Schmitt, E. Cuevas,D. Savoie,W. Graustein,K.
Luria, M., and H. Sievering,Heterogeneousand homogeneousoxidaTurkenian, A. Volz-Thomas, S. Oltmans, H. Levy, and A. Diaz,
tion of SO2 in the remote marine atmosphere,Atmos.Environ., 25,
Temporal variability of ozone and aerosolsin the free troposphere
1489-1496, 1991.
over the easternNorth Atlantic, Geophys.Res.Lett., 22, 2925-2928,
1995.
Maahs, H. G., Kineticsand mechanismsof the oxidationof S(IV) by
ozone in aqueoussolutionwith particular referenceto SO2 conver- Pye, K., Aeolian Dust Particles and Dust Deposits,Academic, San
sion in nonurbanclouds,J. Geophys.Res.,88, 10,721-10,732,1983.
Diego, Calif., 1987.
Malm, W. C., J. F. Sisler, D. Huffman, R. A. Eldred, and T. A. Cahill,
Ross, H. B., and K. J. Noone, A numerical investigationof the destructionof peroxyradical by Cu ion catalyzedreactionson aerosol
Spatial and seasonaltrends in particle concentrationand optical
extinction in the United States, J. Geophys.Res., 99, 1347-1370,
particles,J. Atmos Chem., 12, 121-136, 1991.
1994.
Savoie,D. L., and J. M. Prospero,Particle size distributionof nitrate
Mamane, Y., and J. Gottlieb, Heterogeneous reactionson minerals
and sulfatein the marine atmosphere,Geophys.Res.Lett., 9, 12071210, 1982.
with sulfur and nitrogen oxides,J. Aerosol.Sci., 20, 303-311, 1989.
Mamane, Y., E. Ganor, and A. Donagi, Aerosol compositionof urban Savoie,D. L., J. M. Prospero,and E. S. Saltzman,Non-sea-saltsulfate
and nitrate in trade wind aerosolsat Barbados:Evidence for long
and desert origin in the Eastern Mediterranean, Water Air Soil
Pollut., 14, 29-43, 1980.
range transport,J. Geophys.Res., 94, 5069-5080, 1989.
Matthijsen,J., and D. L. Sedlak,Cloud model experimentsof the effect Savoie, D. L., J. M. Prospero, and R. T. Nees, Nitrate, non-sea-salt
sulfate, and mineral aerosol over the northwestern Indian Ocean, J.
of iron and copper on troposphericozone under marine and continental conditions,Meteorol.Atmos. Phys.,57, 43-60, 1995.
Geophys.Res.,92, 933-942, 1987.
Mozurkewich,M., and J. G. Calvert, Reaction probabilityof N205 on Schutz, L., Sahara dust transport over the North Atlantic OceanModel calculations and Measurements
in Saharan Dust: Mobilisaaqueousaerosols,J. Geophys.Res.,93, 15,889-15,896, 1988.
Mozurkewich,M., P. H. McMur•, A. Gupta, and J. G. Calvert, Mass
tion, Tranport,edited by C. Morales, pp. 267-277, John Wiley, New
York, 1979.
accomodationcoefficientfor HO2 radicalson aqueousparticles,J.
Geophys.Res., 92, 4163-4170, 1987.
Schwartz,S. E., Factorsgoverningdry depositionof gasesto surface
Mukai, H., Y. Ambe, K. Shibata, T. Muku, K. Takeshita, T. Fukuma,
water in PrecipitationScavenging
and Atmosphere-Surface
Exchange,
J. Takahashi,and S. Mizota, Long term variation of chemicalcomvol. 2, edited by S. E. Schwartzand W. G. N. Slinn, pp. 789-801,
position of atmosphericaerosol on the Oki Islandsin the sea of
Hemisphere,Washington,D.C., 1992.
Sehmel, G. A., Particle and gas dry deposition: A review, Atmos.
Japan,Atmos. Environ., 24, 1379-1390, 1990.
Newell, R. E., J. W. Kidson, D. G. Vincent, and G. L. Boer, The
Environ, 14, 983-1011, 1980.
GeneralCirculationof the TropicalAtmosphereand h•teractionswith Shaw, G. E., Considerationson the origin and propertiesof the ant•*• ......
o• •'-,' Geophys.,17, 1983-1998, 1979.
Extratropicall atlt,tg.... nl 9 [lIT Pro•c Oqmhr•ggo[•-•
•0•4
Nishikawa,M., and S. Kanamori, Chemicalcompositionof Kosa aero- Sheehy,D., A Perspectiveon desertificationof grazinglandand ecosol (yellow sand dust) collectedin Japan,Anal. Sci., 7, 1127-1130,
sytemsin North China,Ambio, 21,303-307, 1992.
1991a.
Shettle, E. P., Optical and radiative properties of a desert aerosol
Nishikawa, M., S. Kanamori, N. Kanomori, and T. Mizoguchi, Kosa
model in Proceedings
of the Symposiumon Radiationin theAtmo-
DENTENER
ET AL.: MINERAL
AEROSOL
ON A REACTIVE
SURFACE
22,889
from a numerical simulation,Proceedingsof the Third International
aerosolconference,Kyoto, Japan, pp. 1086-1089, edited by S. Masuda and K. Takahashi, Pergamon,Tarrytown, N.Y., 1990.
Sievering,H., J. Boatman,E. Gorman, Y. Kim, L. Anderson, G. Ennis,
M. Luria, and S. Pandis, Removal of sulfur from the marine boundWesely, M. L., D. R. Cooke, and R. M. Williams, Field measurement
of small ozone fluxesto snow,wet bare soil and lake water, Boundary
ary layer by ozone oxidation in seasaltaerosols,Nature, 360, 571573, 1992.
Layer Meteorol., 15, 361-373, 1981.
Smith, D. M., W. F. Welch, J. A. Jassim,A. R. Chugtai, and D. H.
Westphal, D. L., O. B. Toon, and T. N. Carlson,A two-dimensional
numerical investigationof the dynamicsand microphysicsof SaStedman,Soot-ozonereactionkinetics:Spectroscopic
and gravimetharan dust storms,J. Geophys.Res., 92, 3027-3049, 1987.
ric studies,/tppl. Spectr.,42, 1473-1481, 1988.
Stephens,S., M. J. Rossi, and D. M. Golden, The heterogeneous Westphal, D. L., O. B. Toon, and T. N. Carlson, A case study of
reaction of ozone on carboneceous surfaces, Int. J. Chem. Kin., 18,
mobilizationand transportof Saharandust,J. Atmos. Sci., 45, 2145-
sphere,edited by G. Fiocco, pp. 74-77, A. Deepak, Hampton, Va.,
1984.
1149, 1986.
Stocker, D. W., M.P.
2175, 1988.
Burchardt, M. Plooster, and D. H. Stedman,
Conineouseddy-correlationdry depositionflux measurements,
Final
Rep. WesternEnerg. Supply and Transmissions
Assoc., Rosmead,
Calif., 1987.
Winchester, J. W., and M. X. Wang, Acid-base balance in aerosol
componentsof the Asia-Pacificregion, Tellus,4lB, 323-337, 1989.
Wolff, G. T., On the nature of nitrate in coarse continental aerosols,
Atmos. Environ., 18, 977-981, 1984.
Tegen, I., and I. Fung, Modeling of mineral dust in the atmosphere: Worsnop, D. R., M. S. Zahniser, C. E. Kolb, J. A. Gardner, L. R.
Watson, J. M. van Doren, J. T. Jayne and P. Davidovits, The temSources, transport and optical thickness,J. Geophys.Res., 99,
22,897-22,914, 1994.
perature dependenceof massaccomodationof SO2 and H202 on
aqueoussurfaces,J. Phys.Chem., 93, 1159-1172, 1989.
Tegen, I., and I. Fung, Contributionto the atmosphericmineral aerosol load from land surface modification, J. Geophys.Res., 100, Zhang, X., R. Arimoto, Z. An, T. Chen, G. Zhang, G. Zhu, and X.
18,707-18,726, 1995.
Wang, Atmospherictrace elementsover sourceregionsfor chinese
dust: Concentrations,sourcesand atmosphericdepositionon the
Tegen I., A. A. Lacis,and I. Fung, The influenceon climate forcingof
mineral aerosols from disturbed soils, Nature, 380, 419-422, 1996.
loessplateau, Atmos. Environ., 27, 2051-2067, 1993.
Turner, N. C., S. Rich, and P. Waggoner,Removal of ozoneby soil,J. Zhang, Y., Y. Sunwoo, V. Kotamarthi, and G. Carmichael, PhotoEnviron. Quafit)l,2, 259-264, 1973.
chemicaloxidantprocesses
in the presenceof dust:An evaluationof
the impact of dust on particulate nitrate and ozone formation,
Uematsu, M., R. A. Duce, J. M. Prospero,L. Chen, J. Merrill, R. L.
J. Appl. Meteorol.,33, 1994.
MacDonald,Transportof mineralaerosolfrom Asia overthe North
Zimmermann, P. H., Moguntia: A handy global tracer model in Air
PacificOcean,J. Geophys.Res.,88, 5343-5352, 1983.
Utter, R. G., J. B. Burkholder, C. J. Howard, and A. R. Ravishankara,
Pollution Modeling and its Applications,vol. VI, edited by H. van
Measurement
of the mass accomodation
coefficient
of ozone on
Dop, Plenum, New York, 1988.
aqueoussurfaces,J. Phys.Chem., 96, 4973-4979, 1992.
Van Doren, J. M., L. R. Watson, P. Davidovits,D. R. Worsnop, M. S.
G. R. Carmichael,Department of Chemicaland BiochemicalEngiZahniser,and C. E. Kolb, Temperaturedependenceof the uptake neering, and Center for Global and Regional Environmental Recoefficientsof HNO.•, HC1, and N20 s by water droplets,J. Phys. search,Universityof Iowa, 125A ChemistryBldg.,Iowa City, IA 52240.
Chem., 94, 3265-3269, 1990.
Wang, W., and T. Wang, On the origin and the trend of acid precipation in China, WaterAir Soil Pollut.,in press,1995.
Wang, X., G. Zhu, and X. Shen, Some characteristicsof the aerosol in
Beijing, China J. Atmos. Sci., 19, 211-215, 1990.
Wefers, M., Numerische Simulation der globalcn 3-dimensionalen
Verteilung arider Aerosolpartikelohne bcruecksichtigung
der nasson Deposition,Diplomatheir,Universityof Mainz, (in German),
1990.
Wefcrs, M., and R. Jaenickc, Global 3D distribution of desert aerosol
P. J. Crutzen, Max-Planck-Institut
D-55020 Mainz, Germany.
ffir Chemie, Postfach 3060,
F. J. Dentener and J. Lelieveld, Institute for Marine and Atmo-
spheric Research,Princetonplein5, NL-3584 CC Utrecht, Netherlands.(e-mail: [email protected])
+ 1
Y. z,hm•g, EnvironmcnLa,and Energy SciencesDivision, Battelle
Pacific Northwest Laboratories, Richland, WA 99352.
(ReceivedDecember 14, 1995;revisedMay 28, 1996;
acceptedJune 4, 1996.)