Supplement of Atmos. Chem. Phys., 16, 3743–3760, 2016http://www.atmos-chem-phys.net/16/3743/2016/doi:10.5194/acp-16-3743-2016-supplement© Author(s) 2016. CC Attribution 3.0 License.

Supplement of

Mercury oxidation from bromine chemistry in thefree troposphere over the southeastern US

Sean Coburn et al.

Correspondence to:Rainer Volkamer (rainer.volkamer@colorado.edu)

The copyright of individual parts of the supplement might differ from the CC-BY 3.0 licence.

S1 InstrumentandMeasurementSite

TheinstrumentandmeasurementsiteareidenticaltothatdescribedinCoburnetal.(2011).Only

abriefoverviewwillbegivenhere.Forthedurationofthemeasurementsdiscussedinthisstudy,

aresearch-gradeMAX-DOASinstrumentwaslocatedataUnitedStatesEnvironmentalProtection

Agency(USEPA)facilityinGulfBreeze,FL(30.3N87.2W)andmeasuredfortimeperiodsbetween

May2009andFebruary2011.Thissiteis~10kmsoutheastofPensacola,FL(populationappr.

50,000)and~1kmfromthecoastof theGulfofMexico,whichenables themeasurementof

urbanandmarineairmasses.Thespectrometerandcontrollingelectronicswereset-upinthe

warehouseoftheEPAfacility,whilethetelescopewasmountedonasupportstructureonthe

roofofthewarehouse(~10-12mabovesealevel)connectedviaanopticalfiber.Thetelescope

wasoriented~40°westoftruenorthinordertorealizeaclearviewinthelowestelevationangles

tothecoast.Duringoperationthefull180°elevationanglerangeofthetelescopewasutilizedto

enablethecharacterizationofair-massesovertheseawaterlagoontotheNorth,andoverthe

coastalregionoftheGulfofMexicototheSouth.Forthepurposesofthisstudy,thenorthviewing

directionwillbe considered tominimize changes in the radiative transfer calculationsdue to

azimutheffectsthroughouttheday.

The instrumentusedforthisstudyconsistsofaPrincetonInstrumentsActonSP2300iCzerny-

Turnergrating(500groove/mmwitha300nmblazeangle)spectrometerwithaPIXIS400Bback-

illuminatedCCDdetector(Coburnetal.,2011).Thissetupwasoptimizedtocoverthewavelength

range ~321-488 nm with an optical resolution of ~0.68 nm full-width at half the maximum

(FWHM).Thespectrometeriscoupledtoaweather-resistanttelescope(capableofrotatingthe

elevationangleby180°,50mmf/4optics)viaa10mlong1.7mmdiameterquartzfiber.During

normal field operation this instrumentwas routinely able to realize values of the rootmean

square(RMS)oftheresidualremainingaftertheDOASfittingprocedureontheorder0.9-3x10-4.

Thissystemwasverystable,withlittleneedformaintenance,andwasoperatedremotelyfor

periodsbetweenMay2009andFebruary2011tomeasuremultipletracegases,including:BrO,

IO,nitrogendioxide(NO2),formaldehyde(HCHO),glyoxal(CHOCHO),andtheoxygenmolecule

collisioninducedabsorptionsignal(referredtoasO4).

S2 DOASretrievalsensitivitystudies

S2.1 Retrievalwindow:Severalsensitivitystudieswereperformedtodeterminethemost

suitableanalysissettingsfortheBrOretrieval.Thiswasaccomplishedthroughacomparisonof

bothO3andHCHOdSCDvaluesfromtheBrOfittingwindowwithdSCDspredictedusing

WACCMverticalprofiles.Also,theeffectofusingdifferentO4referencecross-sectionswas

testedwithrespecttotheO4dSCDintheBrOfittingwindow.

WACCMmodeloutputprofilesforO3andHCHOwereusedtoforwardcalculatedSCDsfor

comparisontomeasureddSCDsofO3andHCHOretrievedusingtheBrOfittingwindow.Five

differentBrOanalysissettingwindowsweretested:1)fittingwindow345-359nmwitha2nd

orderpolynomial(encompassestwoBrOabsorptionpeaks–2-bandanalysis);2)fittingwindow

346-359nmwitha2ndorderpolynomial(2-bandanalysis);3)fittingwindow340-359nmwitha

3rdorderpolynomial(3-bandanalysis);4)fittingwindow340-359nmwitha5thorder

polynomial(3-bandanalysis);and5)fittingwindow338-359nmwitha5thorderpolynomial(4-

bandanalysis).Itwasdeterminedthatanalysissetting5)(4-bandanalysiswith5thorder

polynomial)includingaconstrainedintensityoffset(Sect.S2.2)bestrepresentedbothO3and

HCHO.ThesearetheanalysissettingsthatwerethenusedtoassesstheeffectsofdifferentO4

crosssectionsintheBrOfittingwindow:1)Hermans(2002);2)Greenblattetal.(1990);and3)

ThalmanandVolkamer(2013).TheO4dSCDsretrievedfromtheBrOfittingwindowwere

comparedwiththeO4dSCDsretrievedfromanO4optimizedfittingwindowintheUV(353-387

nm);andwhilenoneofthecrosssectionsareabletofullyreproducetheO4dSCDsfromtheO4

optimizedwindow,theThalmanandVolkamer(2013)cross-sectionwasdeemedtobean

improvementoverHermansandGreenblatt/BurkholderinrepresentingO4intheBrOfitting

window.

S2.2 Intensityoffset:AnadditionalparameterthatcanbeutilizedintheDOASretrievalisan

intensityoffset,whichwouldbeusedtohelpaccountforanyinstrumentstraylight.The

instrumentemployedforthisstudywasdesignedtoactivelyminimizespectrometerstraylight

throughtheuseofcut-offfilters(BG3andBG38),aswellasthemethodofbackground

correction.ThebackgroundcorrectionissimilartothatdescribedinWagneretal.(2004)and

utilizesdarkregionsontheCCDdetectortocorrectfordarkcurrentandoffsetnoiseaswellas

straylight.Itwasdeterminedthatstraylightintheinstrumentwasonlyafewpercent(before

correction)inthewavelengthrange330-360nm.Fittinganintensityoffsetshouldonlyaccount

foruncorrectedstraylightandisexpectedtobeontheorderofmagnitudeoftheerrorinthe

backgroundcorrection.ThefittingofthisparametertypicallyhelpsreducetheRMSofthe

fittingroutine,thusimprovinginstrumentsensitivity.However,preliminarystudiesfounda

significanteffectontheretrievedBrOdSCDsdependingonwhetherornotthisparameterwas

includedinthefittingroutine,andthatthiseffectwasmostpronouncedinthenarrowerfitting

windows.Inthemostextremecase(analysiswindow346-359nm),retrievedBrOdSCDs

changedfrom~1x1014moleccm-2withoutfittingtheintensityoffsettovalueslessthanzero

whenanunconstrainedintensityoffsetwasincluded.Inallfittingwindowstested,utilizingan

unconstrainedintensityoffsetresultedinthehighestfitfactorfortheoffsetandlowestvalues

fortheBrOdSCDs,andinsomecasesleadtosignificantlynegative(non-physical)values.In

thesecases,itwasfoundthatperiodsoftimeexistedwhenthefitfactorfortheintensityoffset

wasmuchgreaterthanwhatwasdeterminedtobeareasonablevalueforthisinstrument.For

thisreason,theintensityoffsetwaskeptintheretrieval(tohelpwithRMS),butlimitedtoa

rangedeterminedbytheupperlimitofthisestimatedcorrection(±3x10-3).

BasedonthefindingsoftheoffsetsensitivitytestsandthedSCDcomparisontestspresentedin

thissection,thedSCDsusedfortheBrOinversionarefromthe338-359nmfittingwindow

utilizinga5thorderpolynomial,theconstrainedintensityoffset,andtheThalmanandVolkamer

(2013)O4cross-section.

S3 AerosolRetrieval

Aerosol profiles are determined through an iterative comparison of measured O4 dSCDs

(analyzed in the wavelength window 437-486 nm) with O4 dSCDs calculated from the RTM

McArtim3(Deutschmannetal.,2011)outputsbasedonspecificaerosolextinctionprofiles.This

processisperformedoneachsetofMAX-DOASviewingangles(forthispointforwardreferred

toasascan)ofthecasestudyday(totalof56scans)inordertodetermineindividualaerosol

profiles.Theinitialaerosolprofileusedforeachscandecreasedexponentiallywithaltitudefrom

avalueof0.01km-1at483nm(scaleheightof0.6km).Thiswavelengthischosenforitsproximity

totheO4peakabsorptionstructureat477nmwhileavoidingthefeatureitself,aswellasavoiding

absorptionstructuresfromothertracegases(i.e.NO2).ThisO4fittingwindowwaschosendueto

thebetter general agreement achievedbetweenmeasured and forward calculatedO4dSCDs

(calculatedafteraerosolextinctionprofilesweredetermined)ascomparedwiththeO4retrieval

intheUVrange.ThisissimilartofindingsfromVolkameretal.,(2015),whereitisdetermined

thatusingthe477nmO4bandforderivingaerosolprofilesismuchmorerobustthanusingthe

UV bands due to the increased Rayleigh scattering at shorter wavelengths, which canmask

aerosolextinction.

TheO4verticalprofilesusedforallcalculationsandasinputtotheRTMarebasedontemperature

andpressureprofilesavailablefromNOAA’sESRLRadiosondeDatabaseforlocationsclosetothe

measurementsite,whichareingoodagreementwiththecorrespondingprofilesfromWACCM.

Ineachstepof the iterationthemeasuredO4dSCDsarecomparedtotheforwardcalculated

dSCDsateachelevationangleofthescanbeinganalyzed,andthedifferencesbetweenthese

valuesareusedasinputforoptimizingthemodificationoftheaerosolprofileforthesubsequent

iteration.Forthisstudy,theconvergencelimitissetatapercentdifferenceof5%betweenthe

lowest two elevation angle dSCDs, or if the process reaches 5 iterations without finding

convergencethelastaerosolprofileisused.Thelimitof5iterationsischosenasacompromise

betweenachievingoptimalagreementbetweentheO4dSCDsanddatacomputationtime.For

this case study, the 5% criterion is reached for every sequence. The resulting time series of

aerosolextinctionprofilesareshowninFig.S10(Supplement).

Theaerosolextinctionprofilesat483nmwerescaledtoderiveextinctionat350nmusingthe

relationshipfoundinEq.(S1).

𝜀"#$=𝜀%&"·("#$%&"))*.,# (S1)

whereε350andε483representaerosolextinctioncoefficientsat350and483nm,respectively.An

Angstromexponentof-1.25waschosenasamoderatevaluethatwouldberepresentativeofan

averageatmosphere(Duboviketal.,2002).Theretrievedaerosolprofilesat350nmprovided

inputtotheRTMinordertocalculatetheappropriateweightingfunctionsforBrO.

S4 BoxModelDescription

ThemodelingportionofthisstudyisdesignedtoassesstheconcentrationofBrradicalsavailable

toparticipateinthemercuryoxidationreactionbasedonBrOverticalprofilesprovidedfromthe

MAX-DOASmeasurementsandGEOS-Chem.Othertracegasandatmosphericparameterinputs

to the diurnal steady-state boxmodel (Dix et al., 2013,Wang et al., 2015) aremedian daily

profilesderivedfromtheMAX-DOASmeasurements,WACCM,andGEOS-Chem.Asinglemedian

profile representing the entire day (dailymedian, for eachmodel input) is used rather than

individualprofiles.Fromtheseprofiles,theboxmodelthencalculatesthepartitioningbetween

bromine species throughout the troposphere (including: reactions with other trace gases;

photolysis; and some aqueous phase partitioning and chemistry) in order to derive vertical

profilesoftheBrradical,whicharesubsequentlyusedforcalculatingmercuryoxidationrates.A

summaryofallthereactionsinvolvingmercurycompoundsconsideredintheboxmodelalong

withassociatedratecoefficientscanbefound inTable2.Themodelconceptually followsthe

frameworkofCrawfordetal., (1999),wheremodel inputsare initiatedandallowed to reach

steady-stateoverseveraldays.Intheboxmodel,theBrOandIO(toassesstheimpactofiodine

radical species in the mercury oxidation reactions) profiles are taken from the MAX-DOAS

measurements;however,theGEOS-ChemBrOprofileisalsousedinordertoassesstheimpact

of the differences between these profiles. Other boxmodel inputs are taken as the output

profilesfromtheexternalmodelsutilizedinthisstudyandtheseinclude:temperature,pressure,

andHCHOfromWACCM;andO3andNO2fromGEOS-Chem.AsasensitivitytestWACCMO3and

NO2werealsousedtoassesstheimpactonthemercuryoxidationschemetowardsdifferences

in the vertical distributions of thesemolecules. The resulting vertical profiles of the rate of

mercuryoxidationusingthesetwoprofilesasboxmodelinputsisfoundinFig.S11(Supplement),

whichfollowsthesameformatasFig.7.AerosolsurfaceareameasurementsfromtheTORERO

dataset(Volkameretal.,2015;Wangetal.,2015)areassumedrepresentativeofconditionsin

the marine atmosphere, and therefore used as model inputs for lack of independent

measurements. The vertical distribution of Hg0 from GEOS-Chem was used in all scenarios.

Additionally, photolysis rates for a variety of species, calculated using the Tropospheric

UltravioletandVisible(TUV)Radiationmodel,areincluded.TheTUVmodelwasinitiatedfora

Rayleighatmosphere(aerosolextinction=0),withO3andNO2columnsof380and0.3Dobson

Units(DU),respectively,whicharederivedfromtheaverageverticalprofilesfromWACCM.Since

themedianBrOprofilederivedbytheMAX-DOASmeasurementscloselyresemblestheprofiles

retrievedaroundsolarnoon(seeFig.5),theTUVcalculationsfromthistimeareused.Forthe

modelrunscomparingtheBrOprofilesfromthemeasurementsandGEOS-Chem,onlytheBrO

profileischanged;allotherinputsremainconstant.

Forthedeterminationofthedominantoxidativepathways,thereactionratesforoxidationof

Hg0againstBrandO3arecalculatedasafunctionofaltitudeforthedifferentreactantvertical

profiles, which is important due to atmospheric temperature gradients. This also allows the

assessmentoftherelativecontributionsofthesereactionstotheoverallrateofoxidationasa

functionofaltitude.OxidationbyO3isincludedintheboxmodelduetotheamountofevidence

from laboratory studies indicating that this reaction might play a role in the atmosphere;

althoughpotentiallynotcompletelyinthegasphase.

FollowingWangetal.(2015),theboxmodelisinitiatedundertwodifferentmodestoinvestigate

thesensitivityofoxidationrates,andlikelyproductdistributionstothemechanisticassumptions

aboutmercuryoxidation.ThetwomodesdifferinthescavengingreactionsoftheHgBradduct.

A“traditional”scenarioonlyincludesBrandOHradicalsasscavengers(Holmesetal.,2009),and

a“revised”modeincludesspeciessuggestedbyDibbleetal.,2012(BrO,NO2,andHO2)aswell

asadditionalhalogenspecies(IandIO).Themodelalsotrackstheconcentrationsofallspecies

as a function of altitude,which gives indications for the product distributions of the various

reactions.

References(Supplement)

Coburn,S.,Dix,B.,Sinreich,R.,andVolkamer,R.:TheCUgroundMAX-DOASinstrument:

characterizationofRMSnoiselimitationsandfirstmeasurementsnearPensacola,FLofBrO,

IO,andCHOCHO,Atmos.Meas.Tech.,4,2421-2439,10.5194/amt-4-2421-2011,2011.

Crawford,J.,Davis,D.,Olson,J.,Chen,G.,Liu,S.,Gregory,G.,Barrick,J.,Sachse,G.,Sandholm,

S.,Heikes,B.,Singh,H.,andBlake,D.:AssessmentofuppertroposphericHOxsourcesover

thetropicalPacificbasedonNASAGTE/PEMdata:NeteffectonHOxandother

photochemicalparameters,J.Geophys.Res.-Atmos.,104,16255-16273,

10.1029/1999JD900106,1999.

Deutschmann,T.,Beirle,S.,Friess,U.,Grzegorski,M.,Kern,C.,Kritten,L.,Platt,U.,Prados-

Roman,C.,Pukite,J.,Wagner,T.,Werner,B.,andPfeilsticker,K.:TheMonteCarlo

atmosphericradiativetransfermodelMcArtim:IntroductionandvalidationofJacobiansand

3Dfeatures,J.Quant.Spectrosc.Radiat.Transfer,112,1119-1137,

10.1016/j.jqsrt.2010.12.009,2011.

Dibble,T.S.,Zelie,M.J.,andMao,H.:ThermodynamicsofreactionsofClHgandBrHgradicals

withatmosphericallyabundantfreeradicals,Atmos.Chem.Phys.,12,10271-10279,

10.5194/acp-12-10271-2012,2012.

Dix,B.,Baidar,S.,Bresch,J.F.,Hall,S.R.,Schmidt,K.S.,Wang,S.,andVolkamer,R.:Detection

ofiodinemonoxideinthetropicalfreetroposphere,Proc.Natl.Acad.Sci.U.S.A.,110,

2035-2040,10.1073/pnas.1212386110,2013.

Dubovik,O.,Holben,B.,Eck,T.F.,Smirnov,A.,Kaufman,Y.J.,King,M.D.,Tanre,D.,Slutsker,I.:

Variabilityofabsorptionandopticalpropertiesofkefaerosoltypesobservedinworldwide

locations,J.Atmos.Sci.,59,3,590-608,doi:10.1175/1520-

0469(2002)059<0590:VOAAOP>2.0.CO;2,2002.

Greenblatt,G.,Orlando,J.,Burkholder,J.andRavishankara,A.:Absorption-Measurementsof

Oxygenbetween330nmand1140nm,J.Geophys.Res.-Atmos.,95,18577-18582,

10.1029/JD095iD11p18577,1990.

Hermans,C.,Measurementofabsorptioncrosssectionsandspectroscopicmolecular

parameters:O2anditscollisionalinducedabsorption:

http://www.aeronomie.be/spectrolab/o2,2002.

Wagner,T.,Dix,B.,vonFriedeburg,C.,Frieß,U.,Sanghavi,S.,Sinreich,R.,andPlatt,U.:MAX-

DOASO4measurements:Anewtechniquetoderiveinformationonatmosphericaerosols.

Principlesandinformationcontent,J.Geophys.Res.109,D22205,

doi:10.1029/2004JD004904,2004.al.(2004)

Holmes,C.D.,Jacob,D.J.,Mason,R.P.,andJaffe,D.A.:Sourcesanddepositionofreactive

gaseousmercuryinthemarineatmosphere,Atmos.Environ.,43,2278-2285,

10.1016/j.atmosenv.2009.01.051,2009.

Thalman,R.andVolkamer,R.:Temperaturedependentabsorptioncross-sectionsofO2-O2

collisionpairsbetween340and630nmandatatmosphericallyrelevantpressure,Phys.

Chem.Chem.Phys.,15,15371-15381,10.1039/c3cp50968k,2013.

Volkamer,R.,Baidar,S.,Campos,T.L.,Coburn,S.,DiGangi,J.P.,Dix,B.,Eloranta,E.,

Koenig,T.K.,Morley,B.,Ortega,I.,Pierce,B.R.,Reeves,M.,Sinreich,R.,Wang,S.,

Zondlo,M.A.,andRomashkin,P.A.:AircraftmeasurementsofBrO,IO,glyoxal,NO2,H2O,

O2-O2andaerosolextinctionprofilesinthetropics:Comparisonwithaircraft-/ship-basedin

situandlidarmeasurements,Atmos.Meas.Tech.8,2121-2148,2015.doi:10.5194/amt-8-

2121-2015.

Wang,S.,Schmidt,J.A.,Baidar,S.,Coburn,S.,Dix,B.,Koenig,T.,Apel,E.C.,Bowdalo,D.,

Campos,T.L.,Eloranta,E.,Evans,M.J.,diGangi,J.P.,Zondlo,M.A.,Gao,R.S.,Haggerty,J.A.,

Hall,S.R.,Hornbrook,R.S.,Jacob,D.J.,Morley,B.,Pierce,B.R.,Reeves,M.,Romashkin,P.A.,

terSchure,A.,andVolkamer,R.:Activeandwidespreadhalogenchemistryinthetropical

andsubtropicalfreetroposphere,P.Natl.Acad.Sci.USA,2015.

FigureCaptions

Figure S1: BrO box-AMFs for two elevation angles (25° and 90°, solid and dashed lines

respectively)atdifferentSZAs.Forthesehigherpointingelevationangles,thebox-AMFspeakat

altitudesof2-15km(freetroposphere)forSZAslowerthan70°,whileathigherSZAthebox-AMFs

indicatethatthesensitivitytowardsthefreetroposphereisdecreasing.

FigureS2:Overviewofbox-AMFSforSZAslessthan70°for4elevationangles:a)3.8°;b)10°;c)

25°;andd)90°.

FigureS3:OverviewoftheWACCMBrOverticalprofilesasafunctionofaltitudeandtimeofday

(panela),andthencollapsedintotothecorrespondingpartial(greentrace)andtotal(bluetrace)

VCDsinpanelb.

FigureS4:ExampleresultsfortheinversionofIOfromtheMAX-DOASmeasurements.Panela)

containsthethreea-prioriprofilesusedintheinversion(dashed,coloredlines),thea-posteriori

resultsfromaMAX-DOASscanat~45°SZA(morning)correspondingtothea-prioriprofiles(solid

lines,colorscorrespondwiththea-priori),andthemedianIOprofile(redtrace,wheretheerror

barsreflectthe25thand75thpercentiles).Panelb)showstheaveragingkernelsfortheinversion

using a-priori “Prf1”. Panel c) shows thediurnal variationof the IOVCD for theBL (0-1 km),

troposphere(0-15km),andtotal(0-25km).

FigureS5:OverviewofthesensitivityofSCDRefandthederivedVCDsonthechoiceofreference

spectra/scananda-prioriprofileassumption.Panel(a)containstheSCDRefdeterminedfromboth

forwardcalculationsofthea-prioriprofiles(greytraces)andtheiterativeapproach(bluetraces)

for47differentzenithspectra.Thea-prioricasescorrespondingtothemedianBrOprofilebased

onWACCM (WACCM*1.4) are denotedwith thicker and darker lines. The error bars on the

forwardcalculatedmedianBrOprofilecase(darkgrey)reflectthe±1x1013moleccm-2criteriafor

selectingsuitablereferences.Panel(b)containsthecorrespondingVCDsderivedforoneMAX-

DOAS scan (near solar noon) for each of the references and a-priori combinations. The blue

shadedverticalboxesdenotereferencesthatmeetthecriteriaforbeingusedintheinversion;

andthehorizontalgreybox(panelb)coverstherangeof2.3±0.9x1013moleccm-2,whichfully

capturesallthereferencescontainedwithintheshadedblueregion.

FigureS6:TimeseriesoftheDoFsandinversionRMSforthreedifferentanalysisprocedures:1)

changingreferenceanalysis(referenceisselectedfromeachmeasurementscanthroughoutthe

day,bluetrace);2)fixedreferencewithoutaccountingforanySCDRef(greentrace);and3)fixed

referenceanalysisaccountingforSCDRef(redtrace).IncludedintheplotisSZA(blacktrace)for

reference.

FigureS7:ShowsacomparisonofthemeasuredBrOdSCDsandthedSCDscalculatedfromthea-

posterioriprofiles(inversionusingtheWACCMprofile)forboththeentirecasestudyday(panel

a)andforasmallsubsetofscansbeforesolarnoon(panelb).PanelccontainstheRMSofthe

differencebetweenthemeasuredandcalculateddSCDsforeachscan.

FigureS8:Overviewofverticalprofilesofparametersusedasinputtotheboxmodelutilizedin

this studywhichwere:1) fromWACCM (temperature,pressure,HCHO);2) fromGEOS-Chem

(BrO,O3,NO2,andHg0);3)fromtheTOREROfieldexperiment(totalsurfacearea);and4)derived

duringthisstudy(BrOandIO).

FigureS9:ObservedconcentrationsofselectedtracegasesduringApril2010atthePensacola

MDNsite.Notescalefactorsforsomespeciesgiveninthelegend.

FigureS10:ComparisonofmeasuredO4dSCDsanddSCDcalculatedbasedonthederived

aerosolprofilesfortheentirecasestudyday(panela)andasubsetofscans(panelb).Panelc

containsthediurnalvariationinthederivedaerosolprofiles,andanexampleprofileisfoundin

paneld.

FigureS11:Boxmodelresultsoftherateofmercuryoxidationasafunctionofaltitudefortwo

species:1)ozone(red);and3)bromineradicals(solidblue:MAX-DOAS;dashedpink:GEOS-

Chem)(panela);thecorrespondinglifetimesarefoundinpanelb.Theblackdashedlineat4km

showswheremeasurementsensitivitystartstodropbecauseofthedecreasingamountofBrO

(themeasuredparameter)inthelowerlayersoftheatmosphere.Thisfigureiscomplementary

toFig.7,butusestheWACCM,ratherthanGEOS-Chem,O3andNO2verticalprofilesasinput.

FigureS1

FigureS2

FigureS3

FigureS4

FigureS5

FigureS6

FigureS7

FigureS8

FigureS9

FigureS10

FigureS11