DTIC ADA479811: Factors Influencing the Mesoscale Variations in Marine Stratocumulus Albedo

0.33 MB·English

by  Defense Technical Information Center

Tellus(2007),59B,66–76 (cid:2)C 2007TheAuthors
Journalcompilation(cid:2)C 2007BlackwellMunksgaard
PrintedinSingapore.Allrightsreserved
TELLUS
Factors influencing the mesoscale variations in marine
stratocumulus albedo
By D. A. HEGG1∗, K. NIELSEN2, D. S. COVERT1, H. H. JONSSON2,3 andP. A. DURKEE2,
1DepartmentofAtmosphericSciences,MS351640UniversityofWashington,Seattle,WAUSA;2Departmentof
Meteorology,NavalPostgraduateSchool,Monterey,CA,USA;3CenterforInterdisciplinaryRemotelyPilotedAircraft
Studies,NavalPostgraduateSchool,Monterey,CA,USA
(Manuscriptreceived1February2006;infinalform28September2006)
ABSTRACT
Measurementsofbothhorizontalgradientsandverticalprofilesofaerosols,clouddropletsandthermodynamicparame-
tersinthecloudtoppedmarineboundarylayeroffofcentralCaliforniaarepresented.Theysuggestthat,whileaerosols
canindeedmodulatecloudalbedo,otherparameterssuchasseasurfacetemperaturemaysimilarlyaffectcloudalbedo.
Additionally,theimpactofaerosols,throughsedimentationandprecipitation,oncloudopticaldepthsandthusalbedo
isnotalwaysinaccordwithconventionalexpectationsandcaneitherenhanceordecreasethealbedo,dependingon
ambientconditions.Takentogether,theseresultssuggestthatcurrentestimatesofindirectforcingbyaerosolscouldbe
significantlyinerror.
trievals,foundhighlyvariable(includingbothpositiveandneg-
1. Introduction
ative) correlations between aerosol number concentration and
Numerousstudieshaveproposedindirectaerosolforcing(i.e., retrieved cloud drop effective radius rather than the expected
aerosolforcingthroughmodulationofcloudalbedo)asanim- strongnegativecorrelationoftheTwomeyeffect.ShaoandLiu
portantfactorintheenergybalanceoftheearth-atmospheresys- (2005)haveinterpretedthisintermsofunexpectedcorrelations
tem(e.g.,Twomey,1974,1991;Charlsonetal.,1987;Albrecht, betweencloudgeometricthicknessandaerosolnumberconcen-
1989).Measurementsontheglobalscaledoinfactsuggestthat tration.Similarly,Brenguieretal.(2003)reportnegativecorre-
aerosolsimpactcloudalbedo(Hanetal.,1994)andverylocal lationsbetweencloudgeometricthicknessandaerosolnumber
measurements(ontheorderofafewkm)havedemonstratedthe concentrationfortheACE-2campaign,whichtheyattributepri-
realityoftheaerosol-clouddropletsize–albedoeffectfirstpos- marilytomixingofrelativelyfreshcontinentalairandmarine
tulatedbyTwomeyandnowreferredtoastheTwomeyeffect(cf. air rather than a purely aerosol relationship. Recently, aerosol
Brenguieretal.,2000).Similarly,thereissomeevidenceonlocal indirect forcings (e.g., Ackerman et al., 2004) have also been
scalesthattheclouddropnumber–precipitation–albedorela- suggested that have quite different impacts from conventional
tionshipproposedbyAlbrecht(1989)–Albrechteffect–doesoc- mechanisms.TherationalizationofBrenguieretal.couldhave
cur(Fereketal.,2000).However,itisthemesoscalevariabilityin strongimplicationsforthemagnitudeofaerosolforcing.And,of
cloudalbedothatcontributesmosttotheoverallalbedovariance course,itshouldnotbeforgottenthatnotonlycanaerosolsim-
ofcloudsystems(Rossowetal.,2002),andonthisscaleanum- pactcloudsbuttheconverseisalsotrue,specificallyformarine
beroffactorsotherthanaerosols(e.g.,seasurfacetemperature, stratocumulus(cf.HudsonandFrisbie,1991).Thisreciprocity
SSTasperWyantetal.,1997;drycontinentalairadvectionasper shouldalwaysbeconsideredwheninterpretingobservations.
Brenguieretal.,2003)caninfluencethecloudthermodynamics, Marinestratocumulusdecksplayamajorroleindetermining
possibly sufficiently to result in a net impact on cloud albedo planetary albedo and tend to be located along the eastern pe-
substantiallylessthanexpectedifonlyaerosolswereexerting ripheries of the major oceans (Warren et al., 1988). They will
influence.Observationsyieldalessthanclear-cutpicture.For all experience a mixture of continental and marine air. Along
example,Sekiguichietal.(2003),usingmesoscalesatellitere- these lines, Shao and Liu (2005) found different relationships
betweentheaerosol(non-cloud)opticaldepth–asurrogatefor
∗ aerosolnumberconcentration–andcloudgeometricthickness
Correspondingauthor.
asafunctionofdistancefromthecoastline,withthenear-shore
e-mail:[email protected]
datainagreementwiththefindingsofBrenguieretal.(2003).
DOI:10.1111/j.1600-0889.2006.00231.x
66 Tellus59B(2007),1
Report Documentation Page Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it

does not display a currently valid OMB control number.
1. REPORT DATE 3. DATES COVERED
2007 2. REPORT TYPE 00-00-2007 to 00-00-2007
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Factors Influencing the Mesoscale Variations in Marine Stratocumulus
5b. GRANT NUMBER
Albedo
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
Naval Postgraduate School,Monterey,CA,93943 REPORT NUMBER
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES
14. ABSTRACT

15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE Same as 11
unclassified unclassified unclassified Report (SAR)
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
FACTORS INFLUENCING THE MESOSCALE VARIATIONS 67
These analyses therefore suggest an interesting compensating
influence to aerosol impact on cloud albedo, namely, that the
offshoreflownecessarytobringpollutedairwithhigheraerosol
concentrationsintothemarineenvironment,andimplementthe
TwomeyorAlbrechteffects,willsimultaneouslyadvectdrierair
leadingtosmallergeometriccloudthickness.
Finally,therearemechanismsthatmodulatecloudalbedoin-
dependently of aerosol concentration. Indeed, for the marine
stratocumulusdeckssoimportanttolarge-scaleradiativeforc-
ing (and from which much of the data discussed above is de-
rived), it has long been understood that it is SST more than
perhapsanyothervariablethatmodulatesoverallcloudoptical
depthandalbedo(cf.Kleinetal.,1995;BrethertonandWyant,
1997;Pincusetal.,1997),certainlyforthesynopticscale.(Itis
noteworthythatupuntilthepointwhereSST’saresufficiently
hightoleadtotheSctoCutransition,thecorrelationbetween
SSTandalbedoisgenerallypositive.)Furthermore,Heggetal.
(2004)haveshownthatstratocumulusalbedovariationsonthe
mesoscalecanalsobeattributedtoSSTgradients.Itisentirely
conceivable that aerosol gradients induced by pollution but in
contrary senses to SST gradients might have little impact on
cloudalbedo.
Inlightoftheseissues,wehavegatheredandanalyzedboth
insitudataonaerosolsandcloudmicrophysics,andremotely
retrieved cloud albedo’s and SST’s from the region of exten-
sive marine stratocumulus cloud off of the central California
coast. The measurements were made during July and August
of2004and2005aspartoftheCloudAerosolResearchinthe
MarineAtmosphere(CARMA-IIandCARMA-III)campaigns, Fig.1. (a)GeographiclocationfortheCARMAstudiesand(b)
respectively.Ourgoalistoclarifywhichfactorshaveasignifi- samplingflightplanforthestudy.
cantimpactonthemesoscalevariabilityinmarinestratocumulus
albedo,and,ifpossible,toquantifytherelativeimportanceof
thesefactors. ouspublications(cf.Heggetal.,2002;Wangetal.,2002)but,
perhaps with most relevance to this study, Hegg et al. (2004).
Recapitulatingtheportionsofthesediscussionsrelevanttothe
2. Observationalplan currentanalysis,thePMS/DMTPCASP-100(sizerange:0.12<
d<3.2μm)wastheprimarytoolemployedforaerosolmeasure-
2.1. Experimentalvenue
ment,thePMS/DMTFSSP-100andtheDMTCAPSprobewere
Thegeographiclocationofthemeasurementsreportedhereis usedforclouddropanddrizzledropconcentration,respectively,
showninFig.1a.Theywerethuscentredinoneofthehandfulof andSSTwasmeasuredbymeansofaHeitonicsIRthermome-
persistent(anduniform)marinestratocumulusdecksfoundglob- ter(modelKT19-85).Additionally,forCARMA-III(2005),a
ally.Furthermore,numerousstudiesofthisvenuehavedemon- GerberPVM-100wasavailableformeasurementoftheliquid
strated,primarilyinshiptracks,thatindirectaerosolforcingcan watermixingratio,amorereliabletechniquethanintegrationof
anddoesoccur(e.g.,PlatnickandTwomey,1994;Durkeeetal., theFSSP-100sizedistribution(employedinCARMA-II).While
2000).Thelocationfeaturesbothmesoscalegradientsinaerosol thePVM-100alsoprovidesameasureoftheintegralclouddrop
concentrationandinSST,gradientsthataresometimesorthog- scatteringcrossection,andthusoftheeffectiveradius,thisout-
onal(cf.Heggetal.,2004). putprovedmuchlessreliable,yieldingvaluesincompatiblewith
variousothermeasurements.Hence,foreffectiveradiusdeter-
mination,weusethePVM-100liquidwaterdatacoupledwith
2.2. Theaircraftplatform
theFSSPclouddropnumberconcentration(CDNC)toderive
AlloftheinsitudatadescribedherewereobtainedwiththeCIR- thevolumemeanradiusandthenuseascalingfactorof0.95to
PASTwinOtteraircraftinstrumentationpackage.Mostcompo- relatetheeffectiveradiustothevolumemeanradius(cf.Martin
nentsofthispackagehavebeendescribedinanumberofprevi- etal.,1994;Brenguieretal.,2003).
Tellus59B(2007),1
68 D. A. HEGG ET AL.
Table1. SummaryofthehorizontaltraversesobtainedduringtheCARMA-IIandCARMA-IIIcampaigns
Flight Traverse Numberof Timeof Aerosol–rtot Aerosol–MSE
number Date length(km) verticalprofiles day(UTC) correlation correlation
708 July8,2004 38 2 1716 –0.10 –0.02
709 July9,2004 58 2 1910 0.09 0.2
710 July10,2004 40 2 1718 –0.06 –0.05
712 July12,2004 86 2 1728 0.07 –0.17
713 July13,2004 95 2 1811 –0.13 0.0
721 July21,2004 110 2 1813 –0.24 –0.19
810 August10,2005 91 1 1810 0.19 0.18
816 August16,2005 48 2 2019 –0.68 –0.77
818 August18,2005 121 1 1702 –0.10 –0.13
819 August19,2005 121 0 1728 –0.04 0.08
824 August24,2005 101 1 1848 0.02 0.14
826 August26,2005 121 2 1958 0.41 0.84
pletheverticalthermodynamicstructureoftheMBLalongthe
2.3. Satelliteretrievals
traverse.Aftereachflight,thecloudalbedoalongeachincloud
Satelliteremotesensorswereusedtoretrievethecloudalbedo flightpathwasretrievedfromsatellitedata.Asynopsisofthe
alongtheflighttrackswheretheTwinOttersampledinornear flightplanisshowninFig.1b.
cloud. The sensors used were the AVHRR radiometer in the
NOAA15,16and17satellitesandthestandardspectralradiome-
3. Resultsanddiscussion
teremployedontheGOES-10(West)satellite(channel1,0.63
μm).Bothradiometershavebeendescribedinpreviousstudies 3.1. Horizontalgradients
(e.g., Rao et al., 1999). The absolute accuracy of the albedo’s
InthecourseoftheCARMA-IIcampaign,13researchflights
retrievedbytheseradiometers,particularlythatonGOES-10,is
wereconductedbytheTwinOtterwhile,duringCARMA-III,
uncertainbutweutilizethemhereprimarilytoestablishspatial
16 flights were undertaken. Of these, six during CARMA-II
trendsandforthisonlyrelativeaccuracyisessential.
andsixduringCARMA-IIIcontainedusabledataonmesoscale
(meso-βscaleor∼100km)gradients.Theseflightsaresumma-
2.4. Samplingplan
rizedinTable1.
Theobservedgradients,belowcloud,incloudandfromre-
Thetwoprospectiveforcingfactorsforcloudalbedothatmight
motelyretrievedparametersallshowsubstantialinterflightvari-
reasonablybeconsideredexternal,andthussusceptibletode-
abilityintheirspatialpatterns.Insomeflightsthetrendsinpa-
termination independently of the stratiform cloud deck itself,
rameters are nearly monotonic along the traverses, in others,
areaerosolconcentrationandSST.Mesoscalegradientsinthese
quiteerratic.Mostsignificantly,insomecasesbelowcloudand
variables are indeed present in the operational area and have
in cloud parameters are well correlated with retrieved albedo,
previouslybeenassociatedwithalbedogradients(Heggetal.,
supportingadependenceofthealbedoon,forexample,aerosol
2004). Hence, flight legs were conducted below cloud along
concentrationorSST,whileinothercaseslittlecorrelationwas
pathswhereeitherSSToraerosolgradientsmightbeexpected.
found.ExamplesofthesevariouspatternsareshowninFigs2–5.
Whengradientswereinfactfound,theflightlegswereextended
for about 20 min duration (spatial scale of ∼60 km) directly Itshouldbenotedthatinmostoftheexamplesincorporatedinto
the detailed analysis, HYSPLIT-IV back-trajectories indicated
alongthegradient,ifpossible,butalwaysmoreorlessperpen-
N–NWmarineairfor96hrpriortosampling.Exceptionsare
diculartothemeanwind–evenatthecostofsomeattenuation
Flight708,forwhichthetrajectorybrieflypassedoverland18hr
inthegradient.Thiswasdonesothatthelocalpropertiesalong
priortosampling,andFlight818,whichhadatrajectoryfrom
thegradientreflectedthoseintheoverlyingclouddeck(which
offshoreoftheLosAngelesBasin(effectivelycontinentalair)
wassampledwithanin-cloudflightlegalongpreciselythesame
withatraveltimeof∼20hr.Thiscompareswithaverticalmix-
geographic transect as soon as the below cloud leg was com-
ingtimescale(tomoisturizetheadvectedairtonormalmarine
pleted).Thebasicassumptionimplicitinthisprocedureisthat
themeanMBLverticalmixingtimescaleisshort(∼1hrorless) conditions)of∼24hr.(Stull,1988)
ItisimportanttonotethattheHYSPLIY-IVtrajectorieswere
comparedtothehorizontalmixingtransversetothemeanwind
initializedatseveraldifferentaltitudestotestthesensitivityof
(5–10hr).Usually,asoundingwasconductedfromthesurface
theresults.Typically,theinitialaltitudeswere30m,200mand
towellabovethecloudtopateachendofthetraversestosam-
Tellus59B(2007),1
FACTORS INFLUENCING THE MESOSCALE VARIATIONS 69
withtheaerosolgradient(r=–0.31),andwiththeCDNCgradi-
ent.Flight712alsodisplayscoherencebetweenalbedoandSST
gradients(r=0.96,p=0.003)butwithnosignificantaerosol
gradient present (albedo-aerosol correlation r = –0.018, p =
0.76).Despitethis,theCDNCaswellastheLWCtracktheSST
change,suggestingalargerfractionoftheaerosolareactivated
atthewestendofthetraversewhereCDNC,drizzleandcloud
albedoarehighest(or,possibly,precipitationscavenging).
Other,morecomplex,scenarioscanalsobefound.Forexam-
ple,inFlight708,thebelowcloudgradientsinaerosolandSST
arecoherent(i.e.,changeinthesamesense),leadingtoajoint,
andverystrong,modulationofcloudalbedo(Fig.4).The>60%
increaseinaerosolconcentrationcoupledwiththe2.5◦CSST
warmingalongthetraverseyieldanearlyfactorof3enhance-
mentincloudalbedo.Finally,Flight826showsanalbedotrend
intheoppositesensetotheaerosolgradientwithnocleargradi-
entinSST(Fig.5).Forthisinterestingcase,the50%reduction
incloudalbedotothewestisduelargelytoadecreaseinthe
cloud geometric height (based on vertical profiles at each end
ofthe traverse)aswellas LWCandthus clouddropeffective
radius.
This phenomenology has been previously remarked by
Brenguieretal.(2003)fortheACE-2dataset.Furthermore,Han
etal.(2002),usinganextensivesatellitedatasethaveshownthat
cloudliquidwaterpaths(LWPs)inmarinestratiformcloudsare
commonly negatively correlated with columnar CDNC. How-
ever,theexplanationofferedfortheACE-2observations,con-
current advection of particle-rich but dry continental air, does
notappeartoexplainourobservations,eitherforFlight826in
particular,oringeneral.
Aspreviouslymentioned,HYSPLITmodelbacktrajectories
for the majority of our flights (and all along the traverse for
Flight 826) are entirely marine for 96+ hr prior to sampling.
Nevertheless, this is a far from conclusive criterion and it is
Fig.2. HorizontalgradientsobservedonFlight816for(a)aerosol
worthwhiletoexplorethispossibilityfurther,particularlysince
concentration,(b)SSTand(c)albedo.
quitesmalldifferencesinthebelowcloudmoisturecontentcan
produce significant differences in cloud LWPs. To access this
whatevertheinversionheightwasforaparticularcase.Nosig- possibleexplanationmorefully,wehaveexaminedthecorrela-
nificantdifferencesintrajectorywithinitialaltitudewerefound. tionbetweenaerosolnumberconcentrationandbothtotalwa-
In Fig. 2, data from Flight 816 show significant coherence termixingratio(r )andmoiststaticenergy(MSE)alongthe
tot
(i.e.,similarspatialtrends)betweenthealbedoandthesubcloud belowcloudtraverses.Valuesforthecorrelationsaregivenin
aerosolnumberconcentration.Ontheotherhand,theSSTtrend Table1.FortheACE-2scenariotobeineffect,onewouldexpect
isrelativelysmall(∼1◦Cover∼20km)andintheoppositesense a substantial negative correlationbetween the particle number
towhatwouldbeexpectedtomodulatetheobservedchangein and either of the thermodynamic parameters. However, while
albedo(cf.Heggetal.,2004).Thein-cloudtrends(notshown negativecorrelationsarepresentinabouthalfofthetraverses,
but summarized in Table 4), in accord with this, show a high with one exception they are quite low, explaining 6% or less
degree of correlation of the CDNC with the subcloud aerosol ofthevariancebetweenthecorrelationvariables.Theoneex-
(linearcorrelationcoefficient,r=0.89,fractionalprobabilityof ception,Flight816,whileshowingaverysubstantialnegative
chancecorrelation,p=0.05)buttheliquidwatercontent(LWC) correlation,alsoshowsahighcoherencebetweenthetrendin
hasnocleartrend.Hence,thiscaseprovidesanexampleofthe aerosolnumberconcentrationandcloudalbedo(tobediscussed
firsttypeofindirectaerosolforcing(Twomeyeffect). below),thatis,theACE-2hypothesisisunnecessarysincethere
Incontrasttothis,Flight709(Fig.3)showsanalbedocorre- isnoanomalousrelationshipbetweenaerosolconcentrationand
latedwithSST(r=0.84,p=0.01)butweaklyanti-correlated albedo that requires an explanation. Overall then, the ACE-2
Tellus59B(2007),1
70 D. A. HEGG ET AL.
Fig.3. Horizontalgradientsobservedon
Flight709for(a)aerosolconcentration,
(b)SST,(c)albedoand(d)drizzledrop
concentration.
scenario does not seem applicable to either Flight 826 or our inthesubinversionmixingzone.Ifthesuperinversionairisquite
datasetasawhole,althoughthesensitivityoftheLWP(aswell dry,theresultantenhancedinstabilityandmixingofdryaircan
astheLWCandcloudtopeffectiveradius)toquitesmallchanges dryoutthecloudlayer.Fromeasttowestalongthetraversethe
inthesubcloudwatermixingratio(∼0.1gkg−1)makeitdifficult CDNCincreasesfrom200to350cm−3 leadingtoareduction
tocategoricallyrejectthishypothesis. indrizzletovirtuallyzero(nodrizzledropspresentatall)anda
TheexplanationofferedbyHanetal.fortheirobservations, decreaseinthemeandropsize.Thisleads,asperAckermanetal.,
anegativemodulationofLWPbyprecipitationviadecoupling toenhancedentrainmentofdry,superinversionair,aconsequent
alsodoesnotseemapplicableinourcase.Itiscontrarytomany dryingofthecloudlayer,whichdecreasesingeometricthickness
(though not all) model predictions and is in any case unlikely by45%andinLWCby50%(Fig.6).ThesuperinversionRH
intheshallowcoastalMBLexaminedhere(cf.Brethertonand of36%andthedecreaseinwatervapormixingratioacrossthe
Wyant,1997).Certainly,thereisnoevidenceofdecouplingin inversionof3.7gkg−1 areactuallyquiteclosetotheexample
Flight826or,indeed,inanyofthecasespresentedinthisstudy. fromFIRE-IusedbyAckermanetal.toillustratetheirproposed
Forexample,inFlight826,theMSEisvirtuallyconstantwith mechanism.Theresultingdecreaseincloudalbedoisinaccord
heightbelowtheinversion,havingavalueof310KJkg−1with withthedrying.
aSDof0.3KJkg−1.Similarly r hasaMBLmeanof9.1g Theseexamplesillustratethatthereareanumberoffactors
tot
kg−1withaSDof0.2gkg−1.Thissituationischaracteristicofa that modulate the albedo of marine stratocumulus, that their
well-mixedMBLandistypicalofourdata.DecoupledMBL’s interaction can be complex, and that no single factor, such as
show far more vertical variation in these conserved variables aerosolnumberconcentration,isgenerallypredominant.Quali-
(cf. Martin et al., 1995). It is also conceivable that the east- tativeanalysisofhorizontalgradientsalone,asindeedsuggested
westcontrastissimplyduetomorefavourablethermodynamic bythecaseofFlight826,isunlikelytoleadtoabetterunder-
conditionsattheEasternendofthetraverse,leadingtolarger standingofalbedomodulation.Hence,wenextturntoamore
H and enhanced precipitation. However, as suggested by the quantitative analysis of the data set and examine the vertical
correlationsofr andMSEwithaerosolconcentrationshownin structureoftheclouddeckaswell.
tot
Table1,conditionsareactuallymorefavourabletothewest.The
explanationforthetrendobservedinFlight826appearsratherto 3.2. Verticalprofiles
bealongthelinesrecentlyformulatedbyAckermanetal.(2004)
The cloud decks studied here are quite uniform, having few
who find that the decrease in sedimentation and precipitation
breaksandlittlevariationincloudtopaltitude.Forthesedecks,
due to higher CDNC enhances entrainment through the MBL
withopticaldepths≥5,cloudalbedoisessentiallyafunctionof
inversion.Indeed,suppressionofsedimentationalonewillhave
cloudopticaldepthalone(cf.LacisandHansen,1974)andthe
this effect according to very recent simulations with an LES
opticaldepthcan,inturn,besimplyexpressedas:
model(Bretherton,personalcommunication,2006),essentially
byenhancingtheliquidwateravailableforevaporativecooling τ =2πH Nr2, (1)
Tellus59B(2007),1
FACTORS INFLUENCING THE MESOSCALE VARIATIONS 71
Fig.4. AsinFig.2butforFlight708. Fig.5. AsinFig.2butforFlight826.
whereH isthecloudgeometricthickness(definedhereasthe isevidencethatHandCDNCdonotalwaysvarytogetheraspos-
intervaloverwhichCDNC≥2cm−3),NthemeanCDNCover tulated(i.e.,positivelycorrelated).Tobrieflyreiterate,Brenguier
thatthickness,andrthemeanclouddropradiusforthatthick- etal.(2003)foundthat,fortheACE-2operationsarea,higher
ness (cf. Hobbs, 1993; Twomey, 1991).Note that we use the CDNCwasassociatedwithhighaerosolconcentrationsdueto
clouddropeffectiveradiusasasurrogateforrsincethisisdi- theadvectionofcontinentalpollution.Sincetheadvectedairwas
rectlymeasuredandusedinothercalculations.Thisformulation alsoappreciablydrierthanthebackgroundmarineair,smaller
clearlyshowsthepotentialindependentimpactoffactorsother cloud thickness was associated with the higher CDNC. Simi-
thanCDNC(andthuspresumptivelyaerosolconcentration)on larly,ShaoandLiu(2005)foundthat,forthenearshore(within
thecloudalbedo.Ofcourse,asoriginallyproposedbyTwomey, 1000km)portionoftheCaliforniastratocumulusdeck,therela-
theindirectaerosolforcingassumesaconstantLWCandH,thus tionshipbetweenHandCDNCwasinverse,inagreementwith
imposingacloseandobviousrelationshipbetweenCDNCand theanalysisofBrenguieretal.andcontrarytowhatwouldhave
reff.(thatjointlydetermineLWC).Ontheotherhand,thedepen- been expected from older formulations of indirect forcing by
dence of H on CDNC is much less clear. Numerous previous aerosols.Finally,Hanetal.(2002)foundaninversecorrelation
studieshaveinfactsoughtsuchadependence(usuallyinterms betweentheLWP(afunctionofCDNCandr )andcolumnar
eff
ofprecipitationmodulation)inordertoexaminethedependence CDNCinaboutonethirdoftheirdatasetthough,asdiscussed
ofcloudalbedoonCDNCalone(e.g.,PincusandBaker,1994; above,theprofferedexplanationdoesnotseemapplicabletoour
Heggetal.,1996).However,asperourpreviousdiscussion,there data.
Tellus59B(2007),1
72 D. A. HEGG ET AL.
properly functioning (e.g., Flights 710, 810 and 819 were not
includedduetoalackofgoodalbedoretrievals).Meanparame-
terscharacterizingtheprofilesarereportedinTable2.Inaddition
totheparametersnecessarytoevaluatetheopticaldepthusing
eq.(1),thebelowcloudaerosolconcentrationderivedfromthe
PCASP-100×isgiven,asisthedrizzledropconcentration(drops
withdiametergreaterthan100μm),theretrievedandcalculated
albedo, and the horizontal mean CDNC for a roughly 10 km
portionofthehorizontaltraversecentredontheverticalprofile.
Thislastparameterpermitsatleastanadhocassessmentofhow
representativetheprofilesareofthesurroundingcloudandisde-
rivedsimplytoensurethattheprofilesarenotanomalous.The
linearcorrelationcoefficientbetweentheprofileandhorizontal
CDNCis0.92,supportingtherepresentativenessoftheprofiles.
Preliminarytoexaminingtherelationshipsbetween,andim-
pact of, the parameters shown in Table 2 on cloud albedo, it
isusefultocomparethealbedoderivedfromthemviaeq.(1)
andthealbedoretrievedfromsatelliteradiometers.Areasonable
degreeofclosurebetweentheretrievedandcalculatedalbedo’s
wouldconfirmtheutilityofeq.(1)asausefulformfortherela-
tionshipbetweentheparametersthatdeterminethecloudalbedo.
Aregressionofretrievedontocalculatedalbedousingthedata
inTable2yieldsaregressionequation:
Fig.6. Liquidwaterverticalprofilesforthe(a)eastand(b)westends Aret=(0.61±0.13)Acalc+(5.0±7.7) (2)
ofthehorizontaltraversefromFlight709.
with an R2 value of 0.60 (p < 0.001). The residual variance
associatedwithA issurprisinglyhighandthereisasignificant
ret
Toexplorethisissuewiththecurrentdataset,theparameters biasintheslope(gainbias).Nevertheless,mostofthevariance
ineq.(1)havebeencalculatedfromtheflightslistedinTable1. isexplainedbytheparametersinTable2andsupportstheusage
Sixteenverticalprofilesthroughtheregionalstratocumulusdeck ofeq.(1)asaframeworkfordiscussingthefactorsimpacting
weretakenduringwhichalloftherequisiteinstrumentationwas cloudalbedo.
Table2. ParametersderivedfromtheverticalprofilesobtainedduringtheCARMA-IIandCARMA-IIIcampaigns
Albedo Albedoo Optical Drizzle CDNChorizontal BelowCloud
Flight Location calc. retrieved CDNC(cm−3) reff(μm) H(m) depth dropsa(cm−3) mean Aerosol(cm−3)
708 East 44 20 50 6.3 490 6.1 10 50 200
708 West 81 55 275 6.8 410 33.0 7 250 325
709 East 59 47 300 4.4 300 10.9 7 400 380
709 West 72 44 125 6.3 630 19.6 20 180 180
712 East 76 43 250 5.8 450 23.8 3 250 375
712 West 73 50 400 5.8 240 20.3 1.5 400 380
713 East 25 19 350 3.8 80 2.5 1 400 380
713 West 43 44 350 4.2 150 5.8 1 400 380
721 East 52 36 250 4.2 300 8.3 4.5 325 300
721 West 47 38 325 4.2 190 6.8 2.5 300 300
816 East 67 40 250 5.1 380 15.5 9 150 250
816 West 47 30 130 5.8 250 6.9 20 80 70
818 West 42 22 40 7.4 400 5.5 30 30 50
824 West 61 53 300 4.7 290 12.1 8 300 1000
826 East 58 53 200 5.3 300 10.6 12 225 550
826 West 45 28 350 4.2 165 6.4 1.5 350 650
aClouddropswithdiameter>100μm.
Tellus59B(2007),1
FACTORS INFLUENCING THE MESOSCALE VARIATIONS 73
ExaminationofthecovarianceoftheparametersinTable2re- arestatisticallyindistinguishableforthetwostudies.Whenthe
vealsseveralpointsrelevanttotheissueathand.First,ofcourse, aerosol optical depth (or number concentration) alone is used
thereareanumberofexpectedrelationships.TheCDNChasa asanindependentvariable,ourαvalue(0.16±0.06)issignif-
verysignificantnegativecorrelationwiththedrizzledropcon- icantly lower than that of Shao and Liu but virtually identical
centration(–0.85,p=0.01),CDNCisalsonegativelycorrelated tothatreportedearlierbyNakajimaetal.(2001),0.17,avalue
withtheeffectiveradius(–0.7,p=0.01)butpositivelycorrelated withwhichShaoandLiufeeltheirresultiscomparable.Hence,
withthesubcloudaerosolconcentration(0.55,p=0.05).Onthe overall, our analysis based on in situ measurements compares
otherhand,contrarytoconventionalexpectation,thecloudgeo- quite favourably with the previous satellite work reported and
metricthicknesshasaverysignificantnegativecorrelationwith confirmstheutilityofthemoreindirectapproach.Italso,once
theCDNC(–0.68,p=0.01).Thiscorrelationissimilartothat again,suggeststheimportanceofthecloudgeometricthickness
foundbyBrenguieretal.(2003)fortheACE-2datasetbut,as indeterminingthecloudalbedoandisconsistentwiththelackof
discussedabove,itisunlikelytobeattributabletotheinfluence positivecorrelationbetweenthecloudthicknessandtheaerosol
ofrelativelydry,continentalair.Asmightbeexpectedfromthis, opticaldepth.Thisleadsusbacktothequestionoftheinfluence
thereisalsoanegative,butweaker,correlationbetweenthebe- ofaerosolconcentrationonHasproposedbyAckermanetal.,
lowcloudaerosolconcentrationandthecloudthickness(–0.34, incontrasttothepositivecorrelationexpectedfrommorecon-
p=0.2).Thelarge-scalesatellitedatasetofShaoandLiu(2005), ventionalconceptionsthatconstrainHtoincreasewithCDNC
whichencompassestheCARMAlocale,alsosuggeststhissort orbeindependentofit.
ofrelationship.However,directcomparisonisdifficultbecause Fromthe16profilesshowninTable2,sixmatchedpairsof
ShaoandLiuhadtousetheaerosolopticaldepthasaproxyfor profiles, each from the same traverse but at essentially oppo-
thebelowcloudaerosolconcentration,and,furthermore,could siteends,withsubstantiallydifferentparametervalues,canbe
notusesimultaneousmeasurementsofeventhisparameterwith extracted.Notethat,inprinciple,aseventhpaircouldbetaken
thecloudpropertiessince,ofcourse,thepresenceofcloudpre- fromFlight713.However,thiscasedisplaysnoaerosolgradient,
cludedaerosolopticaldepthretrieval.Indeed,theutilizationof nogradientineitherCDNCordrizzle,andthedrizzleconcen-
climatological mean aerosol optical depths constitutes, in our trationisverylow.ItisclearlyacasewheretheSSTgradient
view,amajoruncertaintyinthefindingsofthisstudy.Inlight isthehighlydominantfactorinmodulationofthealbedoand
ofthesignificanceofitsfindings,wefeelitworthwhiletotest wethereforedonotincludeitinthisanalysis.Thesixmatched
someoftheShaoandLiuresultsusingourinsitudata. pairs,togetherwithvariousparametervaluesnecessarytotestthe
A prominent finding of Shao and Liu, and one that affords applicabilityoftheAckermanetal.mechanismtotheCARMA
anopportunityforcomparisonwithpreviouswork,istherela- data set are given in Table 4. In all cases the relationship be-
tionship between the aerosol optical depth, the cloud geomet- tweentheprecipitationandtheCDNCisasexpected,negative
ricthicknessandtheclouddropeffectiveradius.ShaoandLiu covariance.However,inonlyoneofthecases(816)isthere-
testtherelationshipproposedearlierbyNakajimaetal.(2001), lationship between the CDNC and the cloud thickness as ex-
namely, pectedfromconventionaltheory,thoughinoneother(708)the
cloudthicknesschangesonlymodestly(∼16%).Inthefourre-
reff =στa−αHβ, (3) maining cases, the cloud thickness decreases with increasing
CDNCanddecreasingprecipitation.Ofcourse,insomecases
usingmultiplelinearregressionofthelogarithmofeq.(3).We
(e.g., Flight 709), this is likely due in part to influence of the
dothesamebutnowusingtheinsituconcentrationofaerosol
SST gradient on cloud LWP. However, the phenomenology is
inplaceoftheclimatologicalaerosolopticaldepths.Theresults
alsomoreinaccordwiththeAckermanetal.scenario.Inthis
ofthisanalysisarereportedinTable3,togetherwiththeShao–
regarditshouldbenotedthatthesefourcaseshaveappreciably
Liufindings.Whenusingboththeaerosolopticaldepth(orthe
lowersuper-inversionRH’s(22–39%comparedto52–57%)and
aerosol number concentration in our case) and the cloud geo-
metricthicknessasregressionvariables,thevaluesforαandβ muchlargerdecreasesinthewatervapormixingratioacrossthe
Table3. Regressionparametersfortheanalysisoftherelationshipbetweenclouddropeffectiveradiusand
cloudthickness,andameasureofaerosolconcentration.Symbolsasdefinedineq.(3)
Regressionvariables Study R2 α β Constant[ln(σ)]
lnτa,lnH ShaoandLiu 0.73 –0.07±0.03 0.24±0.02 2.48
lnτa ShaoandLiu 0.44 –0.30±0.03 – 1.70
lnNa,lnH Thisstudy 0.71 –0.10±0.04 0.25±0.06 0.80±0.5
lnNa Thisstudy 0.34 –0.16±0.06 – 2.54±0.34
Tellus59B(2007),1
74 D. A. HEGG ET AL.
Table4. Parametervaluesderivedfromverticalprofilesattheendsoftraversesthatshowtherelationshipbetweenthecloudpropertiesandthe
super-inversionproperties.(Notethattheprimaryvariables,CDNC,HandDrizzlehavemeasurementuncertaintiesof±15%orless,±6%orless
and±80%orless)
Flight Profile CDNC(cm−3) H(m) (cid:7)H/Ha(perpair) Drizzle(cm−3) (cid:7)Albedo(forpair) Super-inversionRH(%) (cid:7)winvb(gkg−1)
708 W 275 410 –0.16 7 .35 57 3
E 50 490 10
709 W 125 630 –0.52 20 .03 31 8
E 300 300 7
712 W 400 240 –0.47 1.5 .07 22 10.8
E 250 450 3.0
721 W 325 190 –0.37 2.5 .02 37 9.7
E 250 300 4.5
816 W 130 250 0.52 20 .10 52 2.5
E 260 380 9
826 W 350 165 –0.45 1.5 –.25 39 7.1
E 200 300 12
aFractionalchangeinHrelativetothechangeinCDNC,thatis,anegativechangemeansachangeintheoppositesensefromthechangeinCDNC.
bThevaluesaremeanvaluesalongeachtraverse.Itshouldbenotedthattherewasverylittlevariationoverthetraverses.
inversion(∼9gkg−1 comparedto∼3gkg−1)thandothetwo Table5. Calculationsofthefractionalchangeinalbedoperchangein
caseswhichdisplayamoreconventionalphenomenology.For CDNCbasedonthedatainTable2.Thelastcolumnisforsucha
example,Flight709hasasuperinversionRHof31%andade- calculationassumingthatcloudthicknessdidnotchange
crease in mixing ratio of 8 g kg−1 across the inversion. This
Flight (cid:7)lnA/(cid:7)ln(CDNC) (cid:7)lnA/(cid:7)ln(CDNC)constantH
leadstoachangeintheLWCprofilequalitativelysimilartothat
predictedbytheAckermanetal.modelforsimilarconditions. 708 0.18 0.20
TheseobservedprofilechangesareshowninFig.6.Whilefirm 709 –0.13 0.12
conclusionscannotbedrawnfromthelimitednumberofcases 712 –0.07 0.15
presentedhere,theseresultsdosuggestthattheAckermanetal. 721 –0.33 0.40
scenario is not rare and could contribute appreciably towards 816 0.19 –
826 –0.29 0.04
contraventionofconventionalaerosol-cloudrelationships.
Thesignificanceofthepotentialdisjointbetweentheincreas-
ingCDNCandthedecreasingcloudthicknessisthattheoppos- impactofentrainmentonalbedoandthusprovideusefulinfor-
ingtendencieswillattenuatetheimpactofaerosolsonthecloud mation.Thefractionalchangeswithandwithouttheassumption
opticaldepthandthusthealbedo.Quantificationofthiseffect, ofconstantthicknessareshowninTable5.Thedecreaseincloud
however,isdifficult;firstsimplybecauseitsoccurrencewillde- thicknessengenderedbyentrainmentofdryairasproposedby
pendonthemesoscalethermodynamicstructureofthesystem, Ackermanetal.hasaverysignificantimpactonthealbedoin
forexample,thesuperinversionhumidity;second,becauseitwill some of these cases, even reversing the albedo gradient from
dependonthemagnitudeoftheaerosolconcentrationascom- thatexpectedfromtheconventionalscenario.However,acau-
paredtotheseotherthermodynamicparameters(cf.Ackerman tionarynoteisinorderhere.RecentworkbyStevensetal.(2003)
etal.).Nevertheless,withafewadhocassumptionswemakea hasshownthatcloudtopentrainmentinmarinestratocumulus
firstattemptatsuchquantification.Basedoneq.(1)andthedata does not always lead to a thinning of the cloud layer if other
inTable2,wefirstcalculatedthefractionalchangeinthealbedo processesaresufficientlystrongtodominateit(e.g.,increasing
perfractionalchangeintheCDNC[((cid:7)lnA/(cid:7)ln(CDNC)].We surfacemoistureflux).Asstatedearlier,anumberofdifferent
thencalculatethesameparameterbutnowartificiallyholding processesaregenerallyatworkinmarinestratocumulusandpre-
thecloudthicknessconstantforthefivecasesinwhichitactu- dictionoftheirneteffectisdifficult.WhattheresultsinTable5
allydecreases,thatis,weassumethattheincreaseintheCDNC suggestissimplythataerosolsthemselvescantendtodecrease
doesnotaffectcloudthickness.(Itisimportanttonotethatthis aswellasincreasecloudalbedo.
procedureisnotaformalpartialderivativebutratheraphysical
assumption.)ObviouslysuchisnotthecaseforeithertheAcker-
manetal.orconventionalscenario,anddoesnotgivetheentire
4. Conclusions
differencetobeexpectedfromthecontrastingmechanismssince
theconventionapproachwillprobablyleadtoincreasedcloud Examination of the horizontal gradients of cloud and sub-
thicknesses.Nevertheless,itwilllikelygivealowerboundforthe cloudpropertiesonthemesoscale,togetherwithsimultaneously
Tellus59B(2007),1