Atmos.Chem.Phys.,16,7389–7409,2016 www.atmos-chem-phys.net/16/7389/2016/ doi:10.5194/acp-16-7389-2016 ©Author(s)2016.CCAttribution3.0License. Biomass-burning impact on CCN number, hygroscopicity and cloud formation during summertime in the eastern Mediterranean AikateriniBougiatioti1,2,3,SpirosBezantakos4,5,IasonasStavroulas3,NikosKalivitis3,PanagiotisKokkalis2,11, GeorgeBiskos6,7,NikolaosMihalopoulos3,7,10,AlexandrosPapayannis2,andAthanasiosNenes1,8,9,10 1SchoolofEarthandAtmosphericSciences,GeorgiaInstituteofTechnology,Atlanta,GA,USA 2LaserRemoteSensingUnit,NationalTechnicalUniversityofAthens,Zografou,Athens,Greece 3ECPL,DepartmentofChemistry,UniversityofCrete,Voutes,71003Heraklion,Greece 4DepartmentofEnvironment,UniversityoftheAegean,Mytilene,81100,Greece 5InstituteofNuclearTechnologyandRadiationProtection,NCSR“Demokritos”,15310Ag.Paraskevi,Athens,Greece 6FacultyofCivilEngineeringandGeosciences,DelftUniversityofTechnology,Delft2728CN,theNetherlands 7EnergyEnvironmentandWaterResearchCenter,TheCyprusInstitute,Nicosia2121,Cyprus 8SchoolofChemical&BiomolecularEngineering,GeorgiaInstituteofTechnology,Atlanta,GA,USA 9InstituteofChemicalEngineeringSciences(ICE-HT),FORTH,Patras,Greece 10IERSD,NationalObservatoryofAthens,P.Penteli15236,Athens,Greece 11IAASARS,NationalObservatoryofAthens,P.Penteli15236,Athens,Greece Correspondenceto:AikateriniBougiatioti([email protected])andAthanasiosNenes([email protected]) Received:14July2015–PublishedinAtmos.Chem.Phys.Discuss.:10August2015 Revised:26February2016–Accepted:24May2016–Published:14June2016 Abstract. This study investigates the concentration, cloud positive matrix factorization (PMF) analysis of the organic condensationnuclei(CCN)activityandhygroscopicproper- ACSM spectra, CCN concentrations follow a similar trend tiesofparticlesinfluencedbybiomassburningintheeastern asthebiomass-burningorganicaerosol(BBOA)component, Mediterraneanandtheirimpactsonclouddropletformation. withtheformerbeingenhancedbetween65and150%(for Air masses sampled were subject to a range of atmospheric supersaturationsrangingbetween0.2and0.7%)withthear- processing (several hours up to 3 days). Values of the hy- rival of the smoke plumes. Using multilinear regression of groscopicityparameter,κ,werederivedfromCCNmeasure- the PMF factors (BBOA, OOA-BB and OOA) and the ob- mentsandaHygroscopicTandemDifferentialMobilityAn- served hygroscopicity parameter, the inferred hygroscopic- alyzer(HTDMA).AnAerosolChemicalSpeciationMonitor ity of the oxygenated organic aerosol components is deter- (ACSM)wasalsousedtodeterminethechemicalcomposi- mined. We find that the transformation of freshly emitted tion and mass concentration of non-refractory components biomass burning (BBOA) to more oxidized organic aerosol ofthesubmicronaerosolfraction.Duringfireevents,thein- (OOA-BB)canresultina2-foldincreaseoftheinferredor- creasedorganiccontent(andlowerinorganicfraction)ofthe ganichygroscopicity;about10%ofthetotalaerosolhygro- aerosol decreases the values of κ, for all particle sizes. Par- scopicity is related to the two biomass-burning components ticlesizessmallerthan80nmexhibitedconsiderablechemi- (BBOAandOOA-BB),whichinturncontributealmost35% caldispersion(wherehygroscopicityvariedupto100%for tothefine-particleorganicwateroftheaerosol.Observation- particles of same size); larger particles, however, exhibited derivedcalculationsoftheclouddropletconcentrationsthat considerablylessdispersionowingtotheeffectsofconden- develop for typical boundary layer cloud conditions sug- sationalgrowthandcloudprocessing.ACSMmeasurements gest that biomass burning increases droplet number, on av- indicate that the bulk composition reflects the hygroscopic- erageby8.5%.Thestronglysublinearresponseofcloudsto ityandchemicalnatureofthelargestparticles(havingadi- biomass-burning (BB) influences is a result of strong com- ameter of ∼100nm at dry conditions) sampled. Based on petition of CCN for water vapor, which results in very low PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. 7390 A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties maximum supersaturation (0.08% on average). Attributing terizetherelativeroleofaerosolnumberandchemicalcom- dropletnumbervariationstothetotalaerosolnumberandthe position(hygroscopicity)variabilitytothepredicteddroplet chemical composition variations shows that the importance numbervariabilityincloudsformedfromBB-influencedair of chemical composition increases with distance, contribut- masses.Theseissuesareimportant,becausethesupersatura- ingupto25%ofthetotaldropletvariability.Therefore,al- tionthatdevelopsincloudsisnotknownbeforehand,noris though BB may strongly elevate CCN numbers, the impact itconstant,butratherastrongfunctionoftheCCNlevelsand ondropletnumberislimitedbywatervaporavailabilityand clouddynamicalforcing(updraftvelocity). depends on the aerosol particle concentration levels associ- Inthecurrentstudywefocusonthehygroscopicity,CCN atedwiththebackground. concentrations and resulting droplet formation characteris- tics (droplet number and cloud supersaturation) associated withairmassesinfluencedbysummertimebiomass-burning events in the eastern Mediterranean. The smoke-laden air 1 Introduction masses sampled were subject to a range of atmospheric processing (several hours up to three days), identified us- Globally, biomass burning (BB) is a major source of at- ing remote sensing techniques (Moderate Resolution Imag- mospheric aerosols (Andreae et al., 2004). In the eastern ing Spectroradiometer, Kaufman and Remer, 1994; Cloud- Mediterranean,uptoone-thirdofthedrysubmicronaerosol Aerosol Lidar with Orthogonal Polarization, Winker et al., mass during the summer period consists of highly oxidized 2009; Mamouri et al., 2012), back-trajectory analysis and organiccompounds(Hildebrandtetal.,2010).DuringJuly– other in situ chemical metrics. Values of the hygroscopic- September, biomass-burning aerosol originates from long- ity parameter, κ, were derived from CCN and Hygroscopic range transport from southern Europe and countries sur- TandemDifferentialMobilityAnalyzer(HTDMA)measure- rounding the Black Sea (Sciare et al., 2008). Bougiati- ments and linked to distinct chemical constituents identi- oti et al. (2014) showed that of the total organic aerosol fied with positive matrix factorization of the chemical con- (OA),about20%isfreshlyemittedbiomass-burningorganic stituents measured with an Aerosol Chemical Speciation aerosol(BBOA),30%isoxidized,processedOAoriginating Monitor (ACSM). Finally, the observations are used to pre- from BBOA (BB-OOA), and the remaining 50% is highly dicttheclouddropletnumberandsupersaturationformedin oxidizedaerosolthatresultsfromextensiveatmosphericage- clouds that develop in each air mass, focusing on the con- ing.Hence,intermsoforganicmass,duringtimeperiodsof tribution of aerosol number and hygroscopicity to the pre- high biomass-burning activity, at least 50% of the aerosol dicteddropletnumbervariability.Thisisoneoftheveryfew canbeattributedtoBBemissions. field studies that use in situ observations to (i) unravel the Aerosol liquid water content (LWC) is a key medium for contributionsofcompositionandaerosolsizetoBB-related atmosphericchemistrythatalsodrivesthepartitioningofsol- CCNdistributionsandtheirimpactsonclouddropletnumber ubleorganicvaporstotheparticlephase(CarltonandTurpin, and (ii) quantify the contributions of biomass-burning con- 2013). LWC is a prime modulator of aerosol direct radia- stituentstoaerosolhygroscopicityandliquidwaterinthere- tiveforcing(e.g.,Pilinisetal.,1995),andbypromotingsec- gion. ondary aerosol formation it can influence aerosol mass and number that impact both the direct and indirect effect of aerosols(Kanakidouetal.,2005). 2 Experimentalmethods Biomass-burningaerosolparticleshavethepotentialtoact as cloud condensation nuclei (CCN), thereby impacting on 2.1 Samplingsiteandperiod cloud properties and climate. Modeling studies suggest that BBisasignificantglobalsourceofCCNnumber(Spracklen The measurements were performed at the Finokalia at- et al., 2011). Laboratory and field studies have shown that mospheric background station (35◦32(cid:48)N, 25◦67(cid:48)E; http:// biomass-burning aerosol is highly hygroscopic and water- finokalia.chemistry.uoc.gr)oftheUniversityofCrete,which soluble,exhibitinguptoabouthalfthewateruptakecapacity is part of the Aerosols, Clouds, and Trace gases Re- of ammonium sulfate (Asa-Awuku et al., 2008; Cerully et search Infrastructure Network (ACTRIS; http://www.actris. al.,2015).Engelhartetal.(2012)foundthatfreshlyemitted net/). More details about the sampling site are provided by BBOAdisplaysabroadrangeofhygroscopicity(κ parame- Mihalopoulos et al. (1997) and Sciare et al. (2003). Al- terfrom0.06to0.6)thatconsiderablyreducesafterjustafew though measurements took place from mid-August to mid- hoursofphotochemicalageing,toaκvalueof0.2±0.1(Pet- November 2012, the focus of our analysis involves the pe- tersandKreidenweis,2007).Fewstudies,however,focuson riods of intense biomass-burning influence from August to the hygroscopicity of ambient BB aerosol as a function of September2012.BBplumessampledwerefresh,originating atmosphericageextendingouttoafewdays.Relativelyfew from the Greek islands and mainland (transport time 6–7h) studiesalsogobeyondCCNtocalculationsofdropletnum- but also from long-range transport from the Balkans (trans- ber(e.g.,Robertsetal.,2003),andevenfewerstudiescharac- port time >1 day) as determined by using Hybrid Single Atmos.Chem.Phys.,16,7389–7409,2016 www.atmos-chem-phys.net/16/7389/2016/ A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties 7391 ParticleLagrangianIntegratedTrajectoryModel(HYSPLIT) back-trajectory analysis as shown in detail in Bougiatioti et al.(2014),combinedwiththehotspots/firedatafromModer- ateResolutionImagingSpectroradiometer(MODIS)/FireIn- formationforResourceManagementSystem(FIRMS;Remy andKaiser,2014). 2.2 Instrumentationandmethodology Chemical composition and mass concentration of non- refractorycomponents(ammonium,sulfate,nitrate,chloride and organics) of the submicron aerosol fraction was pro- vided by an Aerodyne Research Aerosol Chemical Speci- ation Monitor (ACSM; Ng et al., 2011) with a temporal resolution of 30min. More details of the ACSM measure- ments and subsequent analysis can be found in Bougiati- oti et al. (2014). Total absorption measurements provided theblackcarbon(BC)concentrationsbyaseven-wavelength aethalometer (Magee Scientific, AE31). From the BC mea- surementsandusingtheapproachofSandradewietal.(2008) thewoodburningandfossilfuelcontributiontothetotalBC concentrationswerecalculatedusinganabsorptionexponent of1.1forfossilfuelburningand1.86forpurewoodburning. Theaerosolparticle-sizedistributionsfrom9to850nmwere measuredwitha5minresolutionbyacustom-builtscanning Figure1.SchematicofthesetupusedfortheCCNandmixingstate mobilityparticlesizer(SMPS;TROPOS-type,Wiedensohler measurements. et al., 2012) equipped with a condensation particle counter (CPC; TSI model 3772; Stolzenburg and McMurry, 1991). Sample humidity was regulated below the relative humid- al., 2006; Bezantakos et al., 2013) coupled with a conden- ity of 40% with the use of Nafion® dryers in both aerosol sation particle counter (CPC, TSI Model 3772). The RH in andsheathflow,andthemeasurednumbersizedistributions both the aerosol and the sheath flow in DMA-2 was con- werecorrectedfordiffusionalparticlelosses(Kalivitisatal., trolled by proportional-integral-derivative (PID) controllers 2015). towithin±2%accuracy.BothDMAsintheHTDMAsystem A Continuous Flow Stream-wise Thermal Gradient CCN were calibrated with polystyrene latex (PSL) spheres. The Chamber (CFSTGC; Roberts and Nenes, 2005) was used other classified stream was introduced into the CFSTGC to in parallel with a Hygroscopic Tandem Differential Mobil- measuretheCCNactivityofparticles.TheCFSTGCwasop- ityAnalyzer(HTDMA;RaderandMcMurry,1986)tomea- eratedinscanningflowCCNanalysis(SFCA)mode(Moore sure the CCN number, activity and hygroscopicity of ambi- andNenes,2009),inwhichtheflowrateinthegrowthcham- entaerosolforsupersaturated,(0.1–0.7%)andsubsaturated berchangesovertime,whileaconstantstreamwisetemper- conditions(relativehumidity,RH=86%),respectively.The ature difference is applied. This causes supersaturation to whole system, which is illustrated in Fig. 1, sampled air changecontinuously,allowingtherapidandcontinuousmea- withatotalflowrateof1.8Lmin−1.Afterpassingthrougha surementofCCNspectrawithhightemporalresolution.The Nafiondryer(MD-110-12S-2,PermaPureLLC,RH<30%) SFCA cycle used involved first increasing the flow rate lin- the dried particles were selected based on their electrical earlybetweenaminimumflowrate(Q ∼300cm3min−1) min mobility by a Differential Mobility Analyzer (DMA-1; TSI and a maximum flow rate (Q ∼1000cm3min−1) over a max Model 3080; Knutson and Whitby, 1975). The sheath flow ramptimeof60s.TheflowwasmaintainedatQ for10s max and classified aerosol outlet flow of DMA-1 were 10.8 and and then linearly decreased to Q over 60s. Finally, the min 1.8Lmin−1, respectively, while the mobility diameter was flowratewasheldconstantatQ for10sandthescancy- min changedevery6minbetween60,80,100and120nm. clewasrepeated.TheactivateddropletsintheCFSTGCwere TheclassifiedaerosolfromDMA-1wasthensplitintotwo counted and sized at the exit of its growth chamber with an streams. The first stream was passed through a Nafion tube opticalparticlecounter(OPC)thatdetectsdropletsandclas- humidity exchanger where its RH was increased to 86%. sifiestheminto20sizebinswithdiameterrangingfrom0.7 ThesizedistributionoftheRH-conditionedparticleswasde- to10µmevery1s. termined by a second DMA (DMA-2; custom-made DMA The water vapor supersaturations developed in the CF- using a closed-loop sheath flow with RH control; Biskos et STGC during an SFCA cycle were characterized with am- www.atmos-chem-phys.net/16/7389/2016/ Atmos.Chem.Phys.,16,7389–7409,2016 7392 A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties monium sulfate calibration aerosol following the procedure where κ = 4A3 expresses the dependence of κ on d and 27d3s2 p ofMooreandNenes(2009).Inbrief,anammoniumsulfate p s, A= 4Mwσw is the Kelvin parameter, while M , σ and solution was atomized, dried, charge-neutralized and clas- RTρw w w ρ are,respectively,themolarmass,thesurfacetensionand sified by DMA-1. The resulting monodisperse aerosol flow w thedensityofwater.RistheuniversalgasconstantandT is was split between DMA-2 and the CFSTGC, operating in temperature. SFCA mode and with a CFSTGC streamwise temperature In Eqs. (2), (3), s* and κ∗ correspond to the characteris- difference of (cid:49)T =5K. From this setup, we obtain the in- tic critical supersaturation and hygroscopicity parameter of stantaneous concentrations of the classified aerosol and the the monodisperse aerosol, respectively, and correspond to resulting CCN during the SFCA flow cycles. The ratio of the most probable value of the parameters (Cerully et al., CCN to total aerosol number gives the activation ratio, R , a 2011). From Eq. (3), the probability distribution function whichvarieswiththeinstantaneousvolumetricflowrate,Q, for κ,ps(κ), can be derived for the ambient monodisperse intheCFSTGC.UsingdatafrommultipleSFCAflowcycles, aerosol(Cerullyetal.,2011): R isthenfittoasigmoidfunctionthatdependsonQ: a Ra≡ CCCNN =ao+1+(aQ1/−Qa500)−a2, (1) ps(κ)= E1 dRdaκ(κ) =−(cid:18)κC∗2(cid:0)κκ∗(cid:1)C2−(cid:19)12. (4) where a0,a1,a2 and Q50 are constants which describe the 1+(cid:0)κκ∗(cid:1)C2 minimum, maximum, slope and inflection point of the sig- moidal, respectively. The “critical flow rate”, Q50, corre- Analysis of ps(κ) can provide a direct measure of the sponds to the instantaneous flow rate that produces a level chemicalheterogeneityoftheCCNpopulation.Forthis,we of supersaturation, s, required to activate the measured adoptthemetricofchemicaldispersion,σ(κ),introducedby monodisperse aerosol and determined from the size of the Lance(2007)andfurtherdevelopedinCerullyetal.(2011) classifiedaerosolusingKöhlertheory(Mooreetal.,2012a). andLanceetal.(2013): Repeating the procedure for many sizes of classified am- moniumsulfateresultsintheSFCAcalibrationcurve,which 1 gives the supersaturation in the CFSTGC as a function of R(κ−κ∗)2ps(κ)dκ flowrate(i.eQ50vs.s)throughoutanSFCAflowcycle.Ab- σ2(κ)= 0 , (5) soluteuncertaintyofthecalibratedCCNCsupersaturationis R1 ps(κ)dκ estimatedtobe±0.04%(Mooreetal.,2012a,b). 0 Inourinstrumentsetup,R canchangeeitherfromvaria- a tions in the size of the monodisperse aerosol, dp, or the in- whereσ(κ)isthesquarerootofvarianceofκ∗;asthechem- strument supersaturation, s (or flow rate, Q). The indepen- icalheterogeneityoftheCCNincreases,thedistributionofκ dentlyvariedparameterisindicatedhereafterinparentheses broadens,andσ(κ)becomeslargersothattherangeinCCN infrontoftheactivationratio,e.g.,Ra(Q),Ra(s),Ra(dp)for hygroscopicityisgivenbyκ∗±σ(κ). RaasafunctionofQ,s anddp,respectively. Particle water uptake at subsaturated conditions in the AnalysisofRaobtainedforourexperimentalsetupforam- NafiontubehumidityexchangerandDMA-2wasalsoeval- bientparticlessamplesprovidesveryimportantinformation uated by the growth factor measured for the calibration on the activity and chemical mixing state of the CCN. This (NH ) SO . Particle hygroscopic growth at subsaturated 4 2 4 iscarriedoutasfollows.Foreveryparticlesizedp setbythe conditions(gi)isobtained: DMA-1,R (Q),ismeasuredateveryinstantintheCFSTGC a accordingtoEq.(1).Typicallya0(cid:28)a1;giventhatQands g(RH)= dm(RH), (6) arerelatedthroughthecalibration,Ra(Q)datacanbetrans- dp formedtoR (s): a where d (RH) and d are the geometric mean mobility di- E m p Ra(s)= 1+(cid:0)ss∗(cid:1)C, (2) aamtRetHer=s o8f6t%he)assammpelaesdurpeadrtbicyleDsMatAt-h2eahnyddtrhaeteCdPsCta,tean(di.ea.t, where s, s∗ correspond to Q, Q of the monodisperse thedrystateselectedbyDMA-1(RH<30%).Particle-size 50 distributions at 86% RH were inverted using the TDMA fit aerosol.EandCareparametersdeterminedfromfitting.Ac- algorithm (Stolzenburg and McMurry, 1988), which is also cordingtoCerullyetal.(2011),R (s)representsacumula- a capable of distinguishing between internally and externally tivedistributionofcriticalsupersaturationforparticleswith mixedaerosols(e.g.,Bezantakosetal.,2013). drydiameterd ;Köhlertheorycanthenbeappliedtoexpress p HygroscopicitiesdeterminedfromtheCCNmeasurements R (s)intermsofthehygroscopicityparameterR (κ): a a aredifferentiatedbycorrespondingvaluesfromtheHTDMA E R (κ)= , (3) measurementsbyaddingasubscriptCCNorHTDMA(e.g., a 1+(cid:0)κκ∗(cid:1)C/2 κHTDMA, κCCN). κHTDMA is calculated from the HTDMA- Atmos.Chem.Phys.,16,7389–7409,2016 www.atmos-chem-phys.net/16/7389/2016/ A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties 7393 measuredsubsaturatedhygroscopicgrowthfactors: in detail by Bougiatioti et al. (2014). Nevertheless, the or- ganicaerosolderivedfromtheageingofthebiomass-burning κHTDMA=(g(RH)3−1)exp(gR(RHHA)·dp)−1, (7) afielero,sreogla(OrdOleAss-BofBt)hiedBenBtiOfiAeditfowraaslldeevrievnetdshfraodma(sBimouilgairaptiroot-i 100% et al., 2014). A transport time estimate and back-trajectory analysis were conducted with the plume arrival (h) from with A being the Kelvin parameter defined in Eq. (3). The base time graphics with the help of the HYSPLIT model exponential term of this equation (i.e., the Kelvin term) is (http://www.arl.noaa.gov/HYSPLIT_info.php). usedtoaccountforcurvatureeffectsonvaporpressure. MODIS and CALIOP measurements confirm the validity Anaveragevalueofthehygroscopicparameterateachdry of the Bougiatioti et al. (2014) analysis by clearly showing particle size, d , which is representative of the hygroscopic p theorigin,transportpathandcharacteristicsofthebiomass- propertiesoftheentireparticlepopulation,isobtainedasfol- burningplumefromtheChiosfireon18and19August2012. lows: Indeed, in Fig. 2a we show the MODIS true color image gZmax showingtheplumeemergingfromtheChiosfireson18Au- κHTDMA= κ(g(RH))p(g(RH))dg(RH), (8) gust 2012 as obtained during its 09:39UTC overpass over gmin Greece (Kyzirakos et al., 2014). The blue and red lines de- lineatethegroundtrackoftheCALIPSOsatelliteduringits whereκ(g )isobtainedfromthegrowth-factorprobabil- (RH) overpass over Crete several hours later on 19 August 2012 itydistributionusingEq.(5)andp(g )istheprobability (RH) (the first between 00:27 and 00:40 and the second between ofeachgrowthfactor.g ,g aretheminimumandmax- min max 11:34and 11:47UTC); theredstarshows thesamplingsite imumgrowthfactorsobtainedfromthegrowth-factorproba- at Finokalia station. The CALIPSO vertical profiles of the bilitydistributionandrepresenttheminimumandmaximum aerosol backscatter coefficient (in km−1sr−1) at 532 and gwithnon-zeroprobabilityvalue. 1064nm for the two overpasses are shown in Fig. 2b (left- To support the in situ instruments, we used space-borne handside)togetherwiththecorrespondinglinearparticlede- laserremotesensing(lidar)datafromtheCloud-AerosolLi- polarization ratio at 532nm obtained between 00:27:30 and darwithOrthogonalPolarization(CALIOP)(Mamourietal., 00:40UTC (right-hand side). Comparing the midnight and 2009;Winkeretal.,2009)tocharacterizetheplumesemerg- the daytime aerosol backscatter profiles in Fig. 2b, we ob- ing from the fire hot spots. The fire plume originating from serve that the midnight values are 3–4 times lower than the any location can be tracked by HYSPLIT back-trajectory daytime ones for altitudes up to 3km height. In addition, analysis(Bougiatiotietal.,2014),andlidarobservationscan the daytime observations show a discrete aerosol layer be- be used to check the presence of aerosol layers and aerosol low 1.5km. As for the linear particle depolarization ratio it types.Opticalconfirmationofthesmokeplumesisprovided shows a mean value of 19% up to 1.25km height and less byMODISandFIRMSasshowninthesupplementaryma- than6–10%above(1.25–2km). terialofBougiatiotietal.(2014). Finally, we made use of the classification scheme of the CALIPSO data (Omar et al., 2009) to classify the different subtypes of aerosols in the plume captured during its first 3 Resultsanddiscussion overpassoverCreteon19August2012(00:27–00:40UTC). 3.1 Identifyingperiodsofbiomass-burninginfluence This classification scheme, based on the optical and micro- physical properties of the sampled aerosols indeed reveals Bougiatioti et al. (2014) identified the BB events analyzed the presence of a mixture of smoke, polluted dust and ma- herebythetimeevolutionofabsorptionenhancements(BC) rine particles observed below a 3km altitude (black color in the aerosol, which was further verified by FIRMS and for smoke, brown for polluted dust and blue for marine) as back-trajectory analysis. During these events mass spectro- shown in Fig. 2c, within the depicted area between 39◦N, metricbiomass-burningtracers(i.e.,fragmentsm/z=60and 24.1◦E–37◦N,23.4◦E,justNWoftheFinokaliastationand 73) also exhibited elevated levels. Clear biomass-burning along the CALIPSO ground track. According to this clas- contribution was identified by source apportionment using sification, over Crete the presence of polluted dust (mixed positive matrix factorization (PMF) analysis for four dis- withsmokeandmarineaerosols)prevailswithinthemarine tinctevents.TheBBeventsconsideredincludeaseverefire boundary layer, which for Finokalia is approximately 1km event that burned most of the island of Chios (19–21 Au- (Kalivitis et al., 2007), extending up to 0.8–1.2km height. gust), an extensive wildfire at the Dalmatian coast in Croa- This implies that for the 00:27–00:40UTC time slot, the tiaresultinginsmokeplumesthatspreadacrosstheBalkans BB aerosols sampled by the ground-based in situ measure- during the period 28–30 August, and less extensive fires on mentsatFinokaliawouldcontributeless(duetodilution)to the Greek islands of Euboea (3–5 September) and Andros theglobalaerosolmassloadingthanifmeasuredoverwest- (12–13September).AllfireeventsexhibiteddiscreteBBOA ernCrete. profilesdependingonthebiomass-burningfuel,aspresented www.atmos-chem-phys.net/16/7389/2016/ Atmos.Chem.Phys.,16,7389–7409,2016 7394 A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties Figure2.(a)SatellitecompositeviewfromMODISofthefireplumeemergingfromtheislandofChioson18August2012(courtesyof NASA).TheblueandredlinesdelineatethetwogroundtracksoftheCALIPSOsatelliteduringitsoverpassoverCreteon19August2012 between00:27–00:40and11:34–11:47UTC.(b)Verticalprofilesoftheaerosolbackscattercoefficient(inkm−1sr−1)at532and1064nm (left)andlinearparticledepolarizationratioat532nm(right)measuredbyCALIPSOand(c)verticalprofilesoftheaerosolsubtypescaptured byCALIPSOduringitsoverpassoverCrete;themarkedareaislocatedattheNWofFinokaliastation(00:27–00:40UTC). 3.2 PM composition of that of sulfate. Source apportionment clearly shows that 1 these increases are related to BB influences (Bougiatioti et al.,2014).DuringallBBeventsthereisacleardominanceof The average mass concentration for the whole measure- woodburningoverfossilfuelcontributionstoBC.Thewood ment period (mid-August to mid-November 2012), based burningcomponentofBCisalsoprovidedasareference,de- on the ACSM measurements combined with BC from the pictingtheenhancedcontributionofbiomassburningduring aethalometer,was9.2±4.8µgm−3.Thecorrespondingme- thehighlightedevents. dian concentrations for the main aerosol constituents were Based on the size-resolved CCN activity measurements 3.56,1.31,3.03and0.47µgm−3 forsulfate,ammonium,or- and the inferred hygroscopicity parameter κ of the aerosol ganicsandBC,respectively.Figure3representsthetimese- (Eq.3),itisevidentthatthechangesintheorganicsandsul- ries of the major submicron species where it can be seen fate mass fractions will also influence the CCN concentra- that during the fire events the contribution of organics and tions,activationfractionsandhygroscopicity.AstheACSM BCincreasedsubstantially(from34.9to46.5%fororganics provides bulk submicron chemical composition and thus is andfrom6.1to9.5%forBC)withasimultaneousreduction Atmos.Chem.Phys.,16,7389–7409,2016 www.atmos-chem-phys.net/16/7389/2016/ A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties 7395 trationsforthelargerparticlessizesincreaseandfollowthe BBOA trend. This increase was more pronounced, depend- ingontheproximityofthefireandthereforethetraveltime oftheairmasses.Roseetal.(2010)alsoobservedincreases in CCN during a biomass-burning event near the megacity Guangzhou, China, where CCN number concentrations at s=0.068and0.27%,increasedby90and8%,respectively. The same study attributed these changes to increases in the particlesizewhenBBinfluencewaspresent. Of all particle sizes examined, it appears that those hav- ingmobilitydiameterof60nmexhibittheleastvariabilityin Figure 3. Time series concentrations of major PM1 species that terms of CCN number concentration before and during the contributeintheidentificationoftheBBevents.Theshadedareas BBinfluence(Fig.4,opencircles).Theconcentration,how- representthefourconsideredfireevents. ever,significantlyincreasedduringtheBBeventfromCroa- tia(Fig.4b).Thisevent,togetherwithothersofsmallerex- tent,isassociatedwithnewparticleformation(NFP)events. not able to capture any size-dependent chemical composi- The observed frequency of NPF days at Finokalia is close tion, the size-resolved CCN activity measurements are able to 30% (Kalivitis et al., 2015), regardless of the presence toresolvedistinctCCNactivityandmixingstateofthedif- or not of BB-laden air masses. Based on aerosol chemical ferent particle sizes. These aspects are thoroughly investi- composition, it appears that both gaseous sulfuric acid and gatedinthefollowingsections. organiccompoundstakepartinthegrowthofnucleatedpar- ticlestoCCN-relevantsizes.Theseorganiccompoundsthat 3.3 CNandCCNnumberconcentrationsand contribute to the nuclei growth may be of different origins, biomass-burningevents including biogenic emissions, biomass burning and other possible anthropogenic sources from long-range transport Forallfoureventsofbiomass-burning-influencedairmasses (Kalivitisetal.,2015).FromFig.4bitappearsthatwhenthe arriving at Finokalia, the observed aerosol number concen- BB event is combined with such a NPF event within a few tration increased considerably, regardless of size. The in- hours,60nmparticlesarestronglyinfluencedandtheirCCN creasesarequantitativelyexpressedusingaverageddatafrom concentrations increase considerably. The influence on BB at least 6h prior to the arrival time of the BB smoke. For tothehygroscopicityof60nmparticlesandtheothersizesis particlesizesabove100nm,BBincreasedconcentrationsby examined in a subsequent section. A detailed discussion on 65% for the Chios fire, around 50% for the Croatia fire, theseeventsandtheircontributiontoCCNconcentrationsis 88% for the Euboea fire and about 150% for the Andros providedbyKalivitisetal.(2015). fire. Less pronounced increases were seen for the smaller particle sizes. The corresponding impacts on CCN concen- 3.4 CCNactivationfractionsduringfireevents trations for the classified aerosol are shown in Fig. 4 for all fireevents.Concentrationsaregivenatthecharacteristicsu- Asdemonstratedintheprecedingsection,CCNnumbercon- persaturation,s∗,ofthemonodisperseCCNasclassifiedby centrations during the biomass-burning events proportion- DMA-1 (Sect. 2.2). Within each event, s∗ did not vary by ately increased for the larger particle sizes. Figure 5 shows more than 13.6%; therefore most of the variability in CCN the activation fractions (R (Q)) for three of the four parti- a numbercanbeattributedtovariationsinthesizedistribution, cle sizes and the four considered fire events (120nm is not ratherthanshiftsinthechemicalcomposition(i.e.,s∗). shown as it exhibits the same behavior as 100nm). As an Asexpected,smallerparticlesexhibitahighercriticalsu- indicator of BB influence, we use the concentration of the persaturation(Bougiatiotietal.,2011).Duringperiodswith aged BB factor identified in the ACSM spectra (OOA-BB), smoke influence, critical supersaturations tend to increase, asitrepresentstheatmosphericallyprocessedcomponentof indicating that particles associated with BB are less effec- BBOA.Thisfactorischosenasitconstitutesalargerpartof tive CCN compared to those of the background aerosol. To theorganicaerosol(30%)withBBinfluenceanditsageing quantify the direct influence of biomass burning to particle isexpectedtobereflectedintermsofCCNactivity. andCCNnumberconcentrations,westudiedtheconcentra- For all particle sizes, the activation fractions are derived tion of the BBOA component, identified by PMF analysis fromtheasymptoteofthefittingtothesigmoidalfunctionof of the ACSM mass spectra (Bougiatioti et al., 2014). The the R (s) during each supersaturation cycle and represents a BBOA concentration time series depicts the arrival time of the CCN behavior at the highest supersaturations measured thesmokeandtheintensityoftheBBinfluence. (s>0.6%).Figure 5 showsthat eventhough CCNconcen- The data shown in Fig. 4 indicate that during the major- trationsincreaseforparticleslargerthan80nm,theiractiva- ity of the identified biomass-burning events, CCN concen- tion fractions remain more or less stable and very close to www.atmos-chem-phys.net/16/7389/2016/ Atmos.Chem.Phys.,16,7389–7409,2016 7396 A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties Figure4.CCNconcentrationsfortheselectedparticlesizesduringthearrivalofthesmokeplumesfor(a)Chios,(b)Croatia,(c)Euboea and(d)Andros.Theblacksolidlinerepresentsthebiomass-burningcomponentoftheorganicaerosolatthegiventime. unitythroughouttheevents.Thisobservationimpliesthatal- events where the time for transport and ageing is most lim- most all aerosol particles larger than 80nm are CCN active ited(henceleastagedandhygroscopic).Theparticlechem- at supersaturations higher than 0.6%, within uncertainties. ical dispersion retrievals (Sect. 3.5) also support this view. Thisisnotthecasefor60nmparticleswhichhaveactivation The same conclusion is also drawn from the data provided fractionsat0.6%s (andinthecaseoftheChiosfireactiva- byBougiatiotietal.(2011)forthesamesamplingsiteduring tion fractions of 80nm particles as well at 0.4% s) exhibit summertime,andareverifiedbyanalysisofthechemicaldis- thehighestvariability,withratiosapproachingvaluesaslow persionandHTDMAdatashowninfollowingsections.The as40%.ItcanbeseenthatasconcentrationsoftheOOA-BB evolution of the mixing state of each particle size is further starttoincrease,theactivationfractionsof60nmparticlesat investigated by the HTDMA measurements in a subsequent ∼0.6%s starttodiminish.Itthusappearsthatlargerparti- section(Sect.3.6)aswell. cles are mostly internally mixed, as also seen by their high activation ratios, while small particles could be externally 3.5 Hygroscopicityandchemicalheterogeneityduring mixedpopulations.Anindicationoftheheterogeneityofthe thebiomass-burningevents smallerparticlesizescomparedtothelargeronesistheslope ofthesigmoidfittotheRa(s);thesteepertheslope,themore The characteristic hygroscopicity parameters, κ∗, derived homogeneous the population, and given that the 60nm par- from the CCN measurements for all particle sizes and for ticles exhibited the broader slopes, the more heterogeneous thefourselectedfireeventsarepresentedinFig.6.Asaref- theseparticlesare.Thiscanbeexplainedbyasize-dependent erence for the arrival time and magnitude of the event, the chemicalcompositionandthepresenceofapopulationwith concentration of the BBOA factor is also shown in the fig- notablelowerhygroscopicitythatprohibitstheparticlesfrom ure, which has the characteristics of the freshly emitted BB actingasCCNandcanbeattributedtodifferentsourcesand aerosolandisexpectedtohavemoreinfluenceonthehygro- atmospheric processing (coagulation, cloud processing and scopicityoftheparticles.Thesmallerparticleshavethelow- condensationofsecondaryaerosol)thatgenerallytendtoin- est κ values, and hygroscopicity consistently increased CCN ternallymixtheparticles,renderingthemmoreCCNactive. withsize.Thishygroscopicitytrendhasalsobeenobserved Indeed,thelowestactivationfractionsoccurforthestrongest elsewhere(Duseketal.,2010;Cerullyetal.,2011;Levinet Atmos.Chem.Phys.,16,7389–7409,2016 www.atmos-chem-phys.net/16/7389/2016/ A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties 7397 Figure5.Activationfractionsfortheselectedparticlesizesduringthearrivalofthesmokeplumesfor(a)Chios,(b)Croatia,(c)Euboeaand (d)Andros.Thebrownsolidlinerepresentstheprocessedbiomass-burningcomponentoftheorganicaerosol. al.,2012;Paramonovetal.,2013;Liuetal.,2014)andisat- particlessubjecttoatmosphericprocessingwouldbepresent tributedtotheenrichmentinorganicmaterialofsub-100nm inaseparatemodearound120nm(s ∼0.08%).Particles max particles.Basedonthederivedκ valuesforeachparticlesize larger than 100nm are usually more aged than the smaller andwith knowledgeofthe distinctspeciesidentified bythe particles and more immediately associated with BB plumes ACSM(organics,sulfate)andtheirrespectivehygroscopici- and the atmospheric processing they undergo (Kalivitis et ties,thevolumefractionsfororganicsandinorganics(mainly al.,2015).Thehygroscopicityparameterfor100and120nm ammonium sulfate) were estimated for each particle size. It particles are very similar and the fact that the variability in indeedoccursthat60nmparticlesareonaverage89%com- the respective chemical composition is limited may indeed posedoforganics,whiletherespectivevaluesfor80,100and be attributed to cloud processing, while 80nm particles are 120nm particles are 70, 50 and 41%. Most of the accumu- inbetweenthelowestandhighestκ values,anindication CCN lationmodeparticlesresultfromcondensationofsecondary ofsize-dependentchemicalcompositionofcomponentswith sulfates,nitratesandorganicsfromthegasphaseandcoag- differenthygroscopicities. ulation of smaller particles (Seinfeld and Pandis, 2006). In Figure6alsoshowsthatduringthearrivalofthebiomass- order to examine the contribution of constituents from pri- burning-ladenairmasses,κ valuesofallparticlesizesrange marysourcesthatarenotmeasuredbytheACSMtotheac- within0.2–0.3.Thisobservationisconsistentwithvaluesob- cumulation mode particles, we compared the mass derived servedfromchamberexperimentsoffreshandagedbiomass- fromtheACSM+BCandtheintegratedvolumedistribution burningaerosolandinsitustudiesfromthefield.Engelhart fromtheSMPSconvertedtomass.Duringtheexaminedfire et al. (2012) performed a study where 12 different biomass events,theACSM+BCwasonaverage68.6±19.3%ofthe fuels commonly burnt in North American wildfires were SMPS-derivedmass.Thereforethisisanindicationthatnon- used to characterize their respective hygroscopicity. They refractorymaterialneglectedbytheACSMintheaccumula- found that while κ of freshly emitted BBOA prior to pho- tionmodeparticlesprobablyhasasmallinfluenceonparticle tochemical ageing covered a range from 0.06 to 0.6, after hygroscopicity.Accumulationmodeparticlescanalsoresult a few hours of photochemical processing, the variability of from cloud processing. Based on cloud droplet calculations biomass-burning κ values from the different fuels was re- presented in a subsequent section (Sect. 3.8) it appears that duced and hygroscopicity converged to a value of 0.2±0.1 www.atmos-chem-phys.net/16/7389/2016/ Atmos.Chem.Phys.,16,7389–7409,2016 7398 A.Bougiatiotietal.:ImpactofbiomassburningonCCNproperties Figure6.Characteristichygroscopicityparametersoftheselectedparticlesizesfor(a)Chios,(b)Croatia,(c)Euboeaand(d)Andros.The solidlinerepresentsthebiomass-burningcomponentoftheorganicaerosolatthegiventimeandtheshadedareasrepresentthesmokeplume influenceperiod. (Cerully et al., 2011; Engelhart et al., 2012). Based on the Table1.Calculatedchemicaldispersionintermsofσ(κ)/κforthe derivedhygroscopicityparametersforeachparticlessizebe- fourstudiedfireeventsandallmeasuredparticlesizes. foreandduringtheBBinfluence,itoccursthatsmokecauses a relative decrease of κ in the order of 22% for 80nm par- 60nm 80nm 100nm 120nm ticles, 30.6% for 100nm particles and 30.9% for 120nm Chios 0.85±0.14 0.73±0.14 0.60±0.20 0.41±0.16 particles on average for the four events, while κ for 60nm Croatia 0.77±0.18 0.68±0.19 0.44±0.12 0.41±0.10 particlesdeviatebyonly14%. Euboea 0.70±0.20 0.49±0.10 0.32±0.08 0.29±0.06 Andros 0.71±0.10 0.52±0.13 0.34±0.10 0.30±0.06 During the fire events the contribution of organics and BC to the submicron aerosol mass fraction increased sig- nificantlywhilethepresenceofsulfatedeclined.Thisisex- pectedtoinfluencetheCCNactivityofthesampledaerosol thechemicaldispersionσ(κ)ofthehygroscopicityparame- particles as it would cause variations in the inorganic and ter κ, expressed by the standard deviation of kappa around organic mass fractions. It has already been established that themostprobablehygroscopicityκ∗,anditsdependenceon theκ valueofprimaryaerosoldecreasesastheorganicmass particlesize.AsnormaloperationuncertaintiesandtheDMA fraction of aerosol increases (Petters et al., 2009; Engelhart transferfunctioncaninduceabroadeningofR (s)andR (κ) etal.,2012).Withphotochemicalageing,theincreasedoxy- a a andthereforecontributetoσ(κ),theinferredσ(κ)contains genationofthefreshlyemittedBBOAmayinfluencethehy- afairlyconstantinstrumentoffsetandatime-dependentcon- groscopicity of the organic components, but the concurrent stituentthatisrepresentativeoftherealchemicalvariability. increase of the inorganic fraction of the aerosol contributes This offset value, owing to the DMA transfer function and totheobservedincreaseofκ (inorganiccontentvs.age- CCN otherinstrumentlimitshasbeencalculatedtoberoughly0.25 ing). (Cerullyetal.,2011).Table1showsthecalculatedchemical Toexaminetheimpactofatmosphericprocessingandage- dispersion, in terms of σ(κ)/κ, for the four fire events and ing on the composition of the sampled aerosol, we studied the measured particle sizes. It is immediately apparent that Atmos.Chem.Phys.,16,7389–7409,2016 www.atmos-chem-phys.net/16/7389/2016/
Description: