209 pages
Deutsch

DOAS measurements of iodine monoxide from satellite [Elektronische Ressource] = DOAS-Messungen des Spurengases Iodmonoxid vom Satelliten aus / vorgelegt von Anja Schönhardt

-

Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus

Description

DOAS measurements of iodine monoxidefrom satelliteDOAS Messungen des Spurengases Iodmonoxidvom Satelliten ausVom Fachbereich Physik und Elektrotechnikder Universität Bremenzur Erlangung des akademischen GradesDoktor der Naturwissenschaften (Dr.rer.nat)genehmigte Dissertationvorgelegt vonDipl. Phys. Anja SchönhardtBremen, 03. September 2009Dissertation eingereicht am: 03.09.2009Tag des Promotionskolloquiums: 09.10.20091. Gutachter: Prof. Dr. John P. Burrows2. Gutachter: Prof. Dr. Lars KaleschkeFor my parentsAnne and RudolfAbstractAtmospheric columns of the trace gas iodine monoxide, IO, have been investigated by means ofspectroscopic measurements in the visible wavelength range. For this purpose, solar radiation scat-tered and reflected by the Earth’s atmosphere and surface is recorded by satellite instrumentationin nadir viewing geometry. These spectra have been analysed for the absorption signal of the IOvibronic absorption lines. Employing the Sciamachy sensor mounted on the ENVISAT satellite,global observations of IO from space become possible for the first time. The importance of iodinein the atmosphere lies in its high potential for destroying ozone as well as in the formation of newparticles which is initiated by condensable iodine oxides and may impact on Earth’s radiation bud-get at least locally.A major challenge in this work is the smallness of the observed IO optical depths with respect tothe instrument’s detection limit.

Sujets

Informations

Publié par
Publié le 01 janvier 2009
Nombre de lectures 51
Langue Deutsch
Poids de l'ouvrage 29 Mo

ASDO

ASDO

measurementsofiodine

satellitefrom

monoxide

purengasesSdesMessungen

ausSatellitenvom

dmonoIoxid

VomFachbereichPhysikundElektrotechnik
BremenersitätUnivder

zurErlangungdesakademischenGrades
DoktorderNaturwissenschaften(Dr.rer.nat)
Dissertationgenehmigte

Dipl.Phvys.orgelegtAnjavSocnhönhardt

Bremen,03.September2009

Dissertation

gaT

1.

2.

heingereict

am:

quiums:olloPromotionskdes

er:thGutac

er:thGutac

Prof.

Prof.

Dr.

Dr.

John

Lars

.P

03.09.2009

09.10.2009

Burrows

hkKalesce

roFAnne

ymand

stparenRudolf

Abstract

Atmosphericcolumnsofthetracegasiodinemonoxide,IO,havebeeninvestigatedbymeansof
spectroscopicmeasurementsinthevisiblewavelengthrange.Forthispurpose,solarradiationscat-
teredandreflectedbytheEarth’satmosphereandsurfaceisrecordedbysatelliteinstrumentation
innadirviewinggeometry.ThesespectrahavebeenanalysedfortheabsorptionsignaloftheIO
vibronicabsorptionlines.EmployingtheSciamachysensormountedontheENVISATsatellite,
globalobservationsofIOfromspacebecomepossibleforthefirsttime.Theimportanceofiodine
intheatmosphereliesinitshighpotentialfordestroyingozoneaswellasintheformationofnew
particleswhichisinitiatedbycondensableiodineoxidesandmayimpactonEarth’sradiationbud-
getatleastlocally.
AmajorchallengeinthisworkisthesmallnessoftheobservedIOopticaldepthswithrespectto
theinstrument’sdetectionlimit.TheretrievedIOslantcolumnsarethereforeaveragedovercertain
timeperiodsoftypicallyseveralmonths.WidespreadenhancedIOcolumnshavebeendetected
overtheAntarcticregionwithadetailedspatialandtemporaldistribution.Furtherregionswith
positiveIOdetectionaretheEasternPacificupwellingregionandsomeNorthernHemisphericcoast
lines.AdditionaldatasuchastroposphericBrOdistributions,iceconcentrations,phytoplankton
amountsanddiatomabundanceshavebeenconsideredinspecificcasesforcomparisonanddiscus-
sionpurposes,addressingthequestionofsourcesofatmosphericiodine,whicharemostprobably
biogenic.Successfulcomparisonandvalidationstudiesprovideconfidenceinthenewlydevelopedsatellite
IOproduct,andmodelcalculationshavebeenconductedtoinvestigatetheamountsofprecursors
necessaryfortheexplanationofobservedIOabundances.Whileinsomeanalyses,thelimitations
ofthesatellitemeasurementshavebeenencountered,thepresentedinvestigationshaveadvanced
theprospectsofremotesensingfromspaceforthedetectionoftheminortracegasIO.

Listpublicationsof

1.Articlesinpeer-reviewedjournals

firstAsauthor:

•Schönhardt,A.,Richter,A.,Wittrock,F.,Kirk,H.,Oetjen,H.,Roscoe,H.K.,andBurrows,
J.P.:Observationsofiodinemonoxidecolumnsfromsatellite,Atmos.Chem.Phys.,8,
2008.637-653,

co-author:As

•dael,Brinksma,M.,FaE.yt,J.,C.,Pinardi,Hermans,G.,VC.,olten,H.,Dirksen,Braak,R.J.,R.,RichVlemmix,ter,A.,T.,ScBerkhout,hönhardt,A.A.,J.vanC.,RoSwozen-art,
D.M.,P.Curier,J.,R.Oetjen,L.,H.,Celarier,WittroE.ck,A.,F.,Cede,Wagner,A.,T.,Knap,WIbrahim,.H.,O.VW.,eefkind,deJ.LeeuP.,w,EskG.,es,H.MoJ.,erman,Al-
laart,M.,Rothe,R.,Piters,A.J.M.,andLevelt,P.F.:The2005and2006DANDELIONS
NO2andaerosolintercomparisoncampaigns,JournalofGeophysicalResearch,113,D16S46,
2008.doi:10.1029/2007JD008808,

•D.,Celarier,Goutail,E.A.,F.,PBrinksma,ommereau,E.J.,J.-P.,Gleason,Lambert,J.F.,J.-C.,Vvaeefkind,nRoJ.P.,ozendael,Cede,M.,A.,Pinardi,Herman,G.,J.R.,WittroIonockv,,
E.,F.,ScChen,hönhardt,C.M.,A.,PRicongetti,hter,T.A.,J.,Ibrahim,Sander,S.O.P.,W.,WBucsela,agner,E.T.,J.,BoWjkenig,ov,M.B.,O.,MounSwt,art,G.,D.PSpinei,.J.,
Volten,H.,Kroon,M.,andLevelt,P.F.:Validationofozonemonitoringinstrumentnitrogen
dioxidecolumns,JournalofGeophysicalResearch,113,D15S15,doi:10.1029/2007JD008908,
2008.

•Hains,J.,Boersma,F.,Kroon,M.,Dirksen,R.,Volten,H.,Swart,D.,Richter,A.,Wittrock,
F.,Schoenhardt,A.,Wagner,T.,Ibrahim,O.,vanRoozendael,M.,Pinardi,G.,Gleason,
J.,Veefkind,P.,andLevelt,P.:TestingandimprovingOMIDOMINOtroposphericNO2
usingobservationsfromtheDANDELIONSandINTEX-Bvalidationcampaigns,Journalof
GeophysicalResearch,accepted8October2009,doi:10.1029/2009JD012399,inpress.

2.Selectedoralandposterpresentations(onlyfirstautor)

Oralpresentationsatconferencesandworkshops:

•A.Schönhardt,A.Richter,F.Wittrock,andJ.P.Burrows:ObservationofIOfromSpaceus-
ingSCIAMACHY,SixthACCENT-TROPOSAT-2Workshopon”Observingtracesubstances
fromspaceandintegratingtheresultswithmodels”inBremen,June2007.
•A.Schönhardt,A.Richter,F.Wittrock,H.Kirk,H.Oetjen,andJ.P.Burrows:Seasonalvari-
ationsofIOaboveAntarcticaobservedinthreeyearsofsatellitedataDPGFrühjahrstagung,
2008.h,MarcDarmstadt,

seminars:externalattationsPresen

•A.Schönhardt,A.Richter,F.Wittrock,andJ.P.Burrows:IodinemonoxideaboveAntarctica
-4yearsofsatelliteobservations,Hamburg,invitedtalkintheZMAWSeminar,April2008.

tations:presenosterP

•A.Schönhardt,F.Wittrock,A.Richter,H.Oetjen,J.P.Burrows,M.VanRoozendael,G.
Pinardi,H.Bergwerff,S.Berkhout,R.vanderHoff,H.Volten,D.Swart,andE.Brinksma,
MAX-DOASmeasurementsoftroposphericNO2fromtheDANDELIONS-IIcampaign,DPG
Springmeeting,Regensburg,Germany,March2007.
•A.Schönhardt,A.Richter,F.Wittrock,J.P.Burrows,Firstobservationsofatmospheric
iodineoxidecolumnsfromsatellite,EGUGeneralAssembly2007,Vienna,Austria,April
2007.•A.Schönhardt,A.Richter,F.Wittrock,H.Kirk,H.Oetjen,andJ.P.Burrows:Satelliteobser-
vationsofiodinemonoxideanditsseasonalcycleaboveAntarctica,IGAC10thInternational
Conference2008,Annecy(France),September2008.
•A.Schönhardt,F.Wittrock,A.Richter,H.Kirk,H.Schultei.d.B,andJ.P.Burrows:The
influenceofscatteringandabsorptionprocessesinseawateronatmosphericradiation-results
fromship-borneDOASmeasurements,DPGSpringMeeting,Hamburg,März2009.

Contents

ContentsfiguresofListtablesofListMotivationandductionIntrobackgroundScientific11.1Therelevanceofiodineindifferentfields.........................
1.1.1Iodineanditsrelevanceforhumanandanimalhealth.............
1.1.2Radioactiveiodine.................................
1.1.3Therelevanceofiodineforthebiosphere.....................
1.1.4Therelevanceofiodineintheatmosphere....................
1.2IntroductiontoEarth’satmosphere............................
1.3Theimportanceofozone..................................
1.4Halogensintheatmosphere................................
1.4.1Halogensinthestratosphere............................
1.4.2Halogensinthetroposphere............................
1.5Currentstateofatmosphericiodineresearch.......................
1.5.1Sourcesofreactiveiodinecompoundsintheatmosphere............
1.5.2Troposphericiodinechemistryandozonedepletion...............
1.5.3Higheriodineoxidesandparticleformation...................
1.5.4Iodineinthestratosphere.............................
1.5.5Observationsoftroposphericiodineoxides....................
1.6Atmosphericeffectsonradiation.............................
................................absorptionMolecular1.6.1..................................scatteringElastic1.6.2..................................scatteringRaman1.6.31.7Radiativetransferintheatmosphere...........................
1.7.1Descriptionofradiativetransfer..........................
1.7.2TheSCIATRANradiativetransfercode.....................
1.8DifferentialOpticalAbsorptionSpectroscopy......................
1.8.1TheDOASequation................................
1.8.2TheRingeffectreferencespectrum........................
1.8.3TheAirMassFactor................................
1.8.4TheDOASfittingroutine.............................
1.9Descriptionofinstruments.................................
1.9.1ThesatelliteinstrumentSCIAMACHY......................
1.9.2Ground-basedMAX-DOASsystems.......................
1.9.3Additionalsatelliteinstruments..........................

ivivii155667770131415171819112223262721323434353637393043444548494

i

2Developingtheretrievalofiodinemonoxidefromsatellite
2.1Satellitedataconfigurationandselection.........................
2.2TheDOASretrievalofiodinemonoxide.........................
2.2.1ThedevelopedIOstandardfit...........................
2.2.2Fitqualityandconsistency............................
2.3AirmassfactorconsiderationsfortheIOretrieval....................
2.4DetectionlimitforIO...................................
2.5PrecisionandaccuracyoftheIOretrieval........................
2.6ExampleresultsofglobalIOcolumns...........................
.......................................screeningCloud2.72.7.1Cloudscreeningwithanintensitycriterion....................
2.7.2CloudscreeningusingthePMDbasedclassificationscheme..........
2.8Influencingeffectsontheretrieval.............................
2.8.1Investigatedretrievalsettings...........................
2.8.2Retrievalinthe418-438nmwindow.......................
2.9Thechoiceofthebackgroundspectrum.........................
3ObservationsofIOfromsatellite
.....................................ationsobservGlobal3.13.2ObservationsofIOinAntarctica.............................
3.2.1SeasonalvariationofIOinAntarctica......................
3.2.2IOtimeseriesatHalleyStation,Antarctica...................
3.2.3Detailedanalysisinhighertemporalresolution.................
3.3DiscussionofobservationsinAntarctica.........................
3.3.1BrOobservationsandiceconcentrationinAntarctica..............
3.3.2ComparisonofIOwithBrOandseaicemaps..................
3.3.3IOinseaicecoveredareas.............................
3.3.4IOonAntarcticshelficeregionsandthecontinent...............
3.4ObservationsofIOintheEasternPacific.........................
3.5ObservationsofIOontheNorthernHemisphere.....................
3.6ThedifferencebetweentheArcticandtheAntarcticIOobservations.........
3.7NoteontherelevanceoftheretrievedIOamounts...................
4ValidationandcasestudiesofsatelliteIO
4.1Comparisonwithlong-pathDOASmeasurementsatHalley,Antarctica........
4.2ComparisonwithanindependentstudyusingSCIAMACHYdata...........
4.3Comparisonswithground-basedpassiveDOASmeasurements.............
4.4IOinmid-latitudecoastalregions.............................
4.4.1TidaldatafromSHOM..............................
4.4.2CasestudyforlocationMaceHead........................
4.4.3CasestudyforlocationRoscoff..........................
4.4.4Discussionofthetidalanalysis..........................
5ModelingofatmosphericIOwiththeCAABA/MECCAcode
5.1Modellingstudiesofiodinechemistryintheliterature.................
5.2DescriptionoftheCAABA/MECCAmodel.......................
5.3Objectivesandmodelsettings...............................
........................................resultsdelMo5.45.5Proposedmodelextensions................................

ii

533575752676074767677797282838099559798989101501701111111411711221521621129921231731041141141241441145415741251351651

6

7

AnalysingshipbornedatafortheimprovementofDOASretrievals
6.1Motivationforthefollowinganalysis...........................
6.2Instrumentsandmeasurementdetails...........................
6.3Retrievaloftheliquidwaterabsorption.........................
6.4Themixedwatereffect...................................
6.5Retrievalofthewatereffectinsatellitedata.......................

SummaConclusionsandry

reviationsabbofList

Bibliography

159951061261461071

173

177

178

iii

FiguresofList1.1Temperatureprofileoftheatmosphere..........................10
1.2ASolarspectrummeasuredbySciamachyandacloseupofsomeFraunhoferlines.27
1.3Schematicenergybanddiagramofadiatomicmolecule.................30
1.4Absorptioncrosssectionspectrumofiodinemonoxide.................31
1.5ProcessesofelasticscatteringandinelasticRamanscattering.............33
1.6RamanspectrumforN2andO2..............................33
1.81.7InLotensitcationsyspcovectraeredby(top),nadirandthemeasuremenRingeffecttsonsponeectrumday(bottom)..............................4461
1.9SchematicoftheopticalconfigurationofSciamachy.................47
1.101.11SketcSimplifiedhoftskheetchofMAX-DOtheASMAX-DOinstrumenAStlighsetuptpath.......................geometries...............5409
2.1AtypicalspectrummeasuredbySciamachy......................56
2.2ConvolutionprocedureoftheIOspectrumwiththeSciamachyslitfunction....59
2.42.3SpGlobalectraofmaprelevshoanwingttheabsorptionreferencecrossregionsectionsoverandthetheSouthernRingPeffectacific......................6601
2.5TwoexamplefitresultsofthestandardIOfit......................64
2.72.6(a)GlobalTypicalmapoNOf2thefitRingresult.effect(b)fitGlobalfactordistribution..........................ofNO2...............6665
2.92.8ComparisonComparisonofofbloblocckkairairmassmassfactorsfactorsatfordifferendifferenttasolarlbedozenithcasesangles......................6688
2.10BlockAMFsfordifferentcasesincludingthecaseofa100%reflectingcloud.....69
2.11BlockAMFsfortworelevantexamplelocations.....................72
2.12GlobalmapshowingtheslantcolumnresultsofthestandardIOfit..........77
2.13(a)Visualimpressionofthegroundsceneand(b)theSZAcorrectedintensity....78
2.14IOresultswithout(a)andwith(b)anintensitycriterionforcloudscreening.....79
2.15MapsofIOresultswith(right)andwithout(left)cloudscreeningusingSPICS...81
2.172.16(a)GlobalIOfitmapsresultofIOwithretrievstronglyedinthestructuredproblematicresidual,418-438(b)poornmfittingwindowofthe..........Ringeffect.8854
2.18SatellitemeasurementsforspectralfittingresultsshowninFig.2.20.........86
2.192.20FitSciamaresultschyforstatethefourfromthegroundmapinscenesFig.num2.18beredwithinfourFig.nu2.19mb.ered.ground............pixels...8867
2.21MapshowingthelargePacificreferenceregion(Version2)...............90
2.22IOslantcolumnamountsfortwoorbitscomparingtworeferencemethods......92
3.1GlobalslantcolumnamountsofIOaveragedfromJune2004untilMay2008.....96
3.2GlobalslantcolumnamountsofIOaveragedfordifferentseasons...........97
3.43.3MapSeasonallyshowingavtheeragedAnslantarctictconcolumntinentamounandtstheofloIOabcationoveoftheHalleySouthernStation.Hemisphere........10990
3.5TimeseriesofSciamachyobservationsaboveHalleyResearchStation,Antarctica..100
3.63.7MonSeasonalthlymapsmapsofofIOIOonslantthecolumnsSouthernontheHemisphereSouthernaveragedHemisphereoverforfouryeachearsyear..........110032
3.8MonthlymapsofBrOverticalcolumns..........................108
v

vi

3.9DailymapsoficeconcentrationretrievedfromAMSR-Emeasurements........110
3.10SeaWIFSmonthlyclimatologiesoftheoceanicchlorophyll-aconcentration......113
3.11IOslantcolumnsovertheEasternPacificregionaveragedover4years........117
3.12TimeseriesofIOslantcolumnsintheEasternPacific.................118
3.13TheflowoftheHumboldtCurrent.............................119
3.14SeaWiFSchlorophyllaconcentrationintheEasternPacific..............119
3.15SeaWIFSmissioncompositeofthechlorophyll-aconcentrationintheoceans.....120
3.16DiatomconcentrationderivedfromSciamachycomparedtoIOamounts......121
3.17LongitudinalaverageofIOvaluesbetween30◦and150◦East.............122
3.18ReferenceregionfortheanalysisofIOresultsontheNorthernHemisphere......123
3.19SeasonalaveragesofIOslantcolumnsontheNorthernHemisphere..........124
3.20TherelativeorientationofcertaincoastlinestotheSciamachygroundpixels...126
4.1ComparisonofSciamachyIOwithLP-DOASmeasurements.............131
4.2FourconsecutivedaysofIOslantcolumnsfromSaiz-Lopezetal.(2007a).......134
4.3ComparisonofretrievalresultsontheSouthernHemispherefromdifferentstudies..136
4.4MapshowingthelocationofNy-ÅlesundonSvalbard..................137
4.5IOresultsfromMAX-DOASobservationsinNy-Ålesund,adaptedfromOetjen(2009)138
4.6SeasonallyaveragedIOresultsfortheareaaboveSpitsbergen,Ny-Ålesund......139
5.1StructureoftheCAABA/MECCAmodelingcode....................149
5.2ExcerptoftheMECCAchemistryshowingallreactionswhichinvolveiodinespecies150
5.3SchemeoftheaqueousphasereactionsrelatedtoiodineasconsideredbyMECCA.151
5.4ModelledvolumemixingratiosofIOforthebaserun..................154
5.5ModelledvolumemixingratiosofIOforfivedifferentscenarios............155
5.6MaximumdailyIOvaluesindependenceoftheappliedemissionrates.........156
6.1ThetrackoftheANT-XXIV/4PolarsterncruisefromPuntaArenastoBremerhaven160
6.2SketchoftheviewinggeometryoftheDOAStelescopeonboardtheresearchvessel..161
6.3Lineofsightversustimeforatypicaldayoftheresearchcampaign..........162
6.4(a)Absoluteand(b)differentialabsorptioncoefficientofpureliquidwater......163
6.5Samplefitresultforthewaterabsorptionstructures...................164
6.6FitfactorsofliquidwaterabsorptionfromPolarsternDOASmeasurements......165
6.7AveragedresidualspectrafordifferentLOStakenduringshipstation.........167
6.8Fitresultshowingtheretrievalofthewatereffect....................169
6.9Fitfactorsoftheextractedmixedwatereffectfordifferentlinesofsight........170
6.10MapshowingtheliquidwaterfitfactorsasretrievedfromGOME-2observations...171
6.11FitfactorofthewatereffectfromSciamachy(a)andGOME-2(b)observations..172

blesaTofList

1.11.21.31.4

2.12.22.32.42.52.62.72.8

4.14.24.3

5.15.25.3

6.16.26.3

OverviewoverthecompositionofEarth’satmosphere.................
OvOverviewerviewoofversometheimpSciamaortantchymainmeasuremenctshannels.ofiodineo.....................xidesintheatmosphere....
WavelengthbandsofthePMDs..............................

OvReleverviewantoverconfigurationthemainandretrievalselectionversionsparametersrelevantforappliedthistothethesis..satellite..........data.....
RetrievalsettingsforthecurrentstandardfitofIO...................
SurfaceDetectionandlimitscloudfortypesSciamaconsideredchyinobservtheationsSPICSofIOforalgorithm.typical..............conditions......
AFittmosphericsettingsforeffectstheconsideredproblematiconlyretrievinalselectedV0.27i.testretriev.....................als.............
StatisticalinformationonthedifferencebetweenV2.54C-AVEandV2.54Dresults..

ExampleofthetidaldataprovidedbySHOM......................
ComparisonofIOcolumnsforhighandlowtideatMaceHead,Ireland.......
ComparisonofaveragedIOamountsforhighandlowtideatRoscoff.........

Settingsforthebasicconditionsvalidforallmodelruns.................
Modelparametersandsettingsasusedinthebaserun..................
Settingsfordifferentprecursoremissionrates......................

Overviewoftheappliedinstrumentsandrelevantmeasurementsettings.......
DOASfitsettingsfortheretrievalofliquidwaterabsorption..............
RetrievalsettingsfortheDOASfittoextractthemixedwatereffect..........

9248484

8585263708384892

141341341

351351451

161361661

vii

IntroMotivationandduction

TheatmosphereofplanetEarthprovidesvitalprerequisitesformanylife-forms.Viceversa,theat-
mosphericcompositionissignificantlydeterminedbytheinteractionswithlife.Earth’satmosphere
istheonlyplaceknownintheuniversewiththepresentspecificmixtureofgases,wheretheamount
ofoxygenismuchhigherthanwouldbeexpectedforasysteminchemicalequilibrium(Wayne,
2000).TheEarthsystemwiththedifferentdomains-atmosphere,biosphere,oceansandland-is
describedbythecontroversialbutinterestingandthought-provokingGaiatheory(Lovelock,1990)
asaregulatoryfeedbacksystemsustainingtheconditionsnecessaryfortheexistenceoflife.
TheimportanceofinterdisciplinaryresearchinthecontextofEarthscienceshasbeengenerally
recognised.Notonlyindividualaspectsneedtobestudiedbutalsotheconnectionsbetweenthe
differentfieldsanddomains.Thechemicalcompositionoftheatmosphere,forinstance,impactson
theEarth’sradiationbudgetcausingatemperatureresponsethroughphysicalabsorption,emission
andscatteringprocesses,e.g.,bygreenhousegasesandaerosolparticles.Thisisaccompaniedby
effectsonfurtherconditionssuchaspressureandwindfields,precipitationandoceancirculations
aswellas,consequently,byimpactsonanimalandplanthealth.Somechemicalcompoundsare
stableandhardlyundergointeractions,whileothersareveryreactiveandmayleadtosubstantial
compositionalchanges.ThecompositionisagaininfluencedbyalldomainsoftheEarthsystem,by
emissionsofbiogeniccompounds,bydepositionofsubstancestolandandoceans,bychemicaland
phasetransformationsandbyinnumerableotherprocesses,eachofwhichformsonlyasmallpart
cycles.largerofIodinespeciesarerelevantinmanyrespectsandrepresentagoodexampleofamultidisciplinary
field.Duetomanyinvolveddomainsandpathways,thecombinationofprocessesisreferredtoas
thebiogeochemicalcyclingofiodine.Insomedomains,theyplayacrucialroleinspiteofrather
lowabundances.Althoughonlytracesofiodineareneeded,itisanessentialelementforvertebrate
lifethroughitsinvolvementinthyroidhormonecomposition.Ithasasecondbiologicalroleby
protectingplantsfromoxidativecelldecay.Asobservedforalgaeandphytoplankton,forexample,
gaseousorganiciodinecompoundsareemittedtotheatmosphereintheseinstances.Oceanicsalt
watersuppliesthebasiciodineatomsfortheseorganisms,andchemicalconversionsoftheconcerned
speciesareinvolvedinallsteps.Iodinebelongstothechemicalfamilyofthehalogensandvarious
iodinespeciesarefoundinseasalt,soils,inplantandanimalorganisms,andalsointheatmosphere.
Afewdecadesago,halogensandtheirchemistryhaveadvancedtoanimportantresearchfocus
duetotheirdestructiveimpactontheSouthPolarozone.Ozone(O3)providesaprotectivelayer
inthestratosphereabsorbingenergeticultraviolet(UV)radiationwhichisharmfulfororganisms
livingontheEarth’ssurface.Catalyticreactioncyclesinvolvingchlorineandbrominefromman-
madechloroflourocarbons(CFCs)andhalonsareresponsibleforthestratosphericozonedestruction
asproposedbyMolinaandRowland(1974)aswellasStolarskiandCicerone(1974)andobserved,

1

IntroductionandMotivation

e.g.,byChubachi(1984)andFarmanetal.(1985).
Inthe1980s,aconnectionbetweenhalogenchemistryandprocessesinthelowertropospheric
layerswasdiscovered.OzonelosseventsintheArcticboundarylayerandlowozoneconcentrations
inthetropicaltropospherewereobservedbutcouldnotbeexplainedorreproducedbyatmospheric
chemistrymodels.Strongozonedepletionevents(ODEs)inthePolarboundarylayerwerefound
tobeaccompaniedbyenhancedamountsofbrominecompounds(Barrieetal.,1988).Bromineand
iodinechemistryhavebeenidentifiedsinceasmissinglinksintheunderstandingoftropospheric
ozonelevels.Troposphericozoneinfluencestheatmosphericoxidisingcapacityandisofessential
importance,e.g.,throughtheproductionoftheOHradical,butbecomesharmfulattoohigh
concentrations(summersmog)andinducesadditionallydirectandindirectradiativeforcing.While
mankindhasalreadycausedseriouschangesoftheatmosphericcomposition,e.g.,intermsofair
pollutionandincreasinggreenhousegases,ahealthyenvironmentisadesireaffectingeveryone
onEarth.Ithasalsobecomeamajorresearchobjectivetobetterunderstandthecurrentstate
oftheEarthsystem,therelevantinteractionsandtheongoingchangesinordertopredictfuture
developmentandpossiblyrestrictfurtherimpactasfaraspossible.
Iodinespecieshavetwomainimplicationsforatmosphericcomposition.Oneofthemisthe
largepotentialofatomiciodineforcatalyticozonedestruction.Asecondaspectdiscriminates
iodinefromtheotherhalogens.Iodineoxidesformcondensablevapoursfromwhichsmallparticles
maybegenerated.Thesecangrowtobecomecloudcondensationnuclei(CCN)andconsequently
influencetheclimatestate.SimilartosulfateaerosolfromDMS(dimethylsulphide)oxidation,
particulateiodinecontributestotheamountofnaturalaerosolsandCCN,probablymainlyin
oceanicregions.Inbothprocesses,iodinemonoxide(IO)playsacentralrole,asimmediateproduct
ofozonedepletionbyiodineatomsandasthestartingpointforparticleformationviahigheriodine
oxides.KnowledgeonthepresentIOamountsaswellastheidentificationandquantificationof
iodinesourcesintheatmospherearethereforeimportant.Severalinorganicandorganicrelease
pathwaysarecurrentlydiscussed,butthequestionhasnotbeenfullysolvedyet.
AsIOformsveryfastfromiodineprecursors,itisagoodindicatorforongoingiodinechemistry.
AlthoughIOabundancesobservedsofararecomparativelylow,theirimpactmaybesubstantial,
owingtothefastconversionsandcatalyticcycles.ThefirstatmosphericmeasurementsofIOhave
onlybecomepossibleabout10yearsago(Alickeetal.,1999;Wittrocketal.,2000).SeveralIO
measurementshavesincethenbeenperformedwithground-basedinstrumentationandballoon-
bornedevicesonacampaignbasis.TheseobservationshaverevealedIOatdifferentlocations,
mainlyatcoastalsitesandafewPolarresearchstations.
SeveralquestionsontheabundancesandsourcesofIOstillremainopen.Theoverallimportance
ofiodineisdifficulttoassessbylocalmeasurementsonly,asthelargescalespatialdistributionisnot
revealedandoftenthecampaigndurationrestrictstheinformationcontentontemporalevolution.
Satelliteobservationsingeneralprovideavaluabletoolfortheextensionoftracegasmeasurements
toamoreglobalscale,andhaveimprovedtheknowledgeofamountsandsourceregionsforseveral
tracespeciessuchasO3,NO2,CO2,HCHO,SO2andothers.Afterlargeandwidespreadamounts
ofbromineoxide(BrO)inthespringtimePolarRegionshadbeenobservedfromsatellitesome
yearsago(Richteretal.,1998;WagnerandPlatt,1998),thequestionwasopeninhowfarIOwould

2

IntroductionandMotivation

distribution.similaraealrevPriortothiswork,IOhadnotbeenobservedfromspacebefore.Oneoftheobjectivesof
thepresentthesisistheretrievalofIOcolumndensitiesfromtheSciamachysatellitesensor.
SciamachyismountedonanEarthorbitingsatelliteandrecordssolarradiationscatteredand
reflectedbytheatmosphereandsurface.Byspectroscopicmeans,theamountsofIOaredetermined
fromabsorptionfeaturesinthevisiblespectralregion.Theappliedretrievaltechniqueisthewell
establishedDifferentialOpticalAbsorptionSpectroscopy(DOAS)method.Thechallengingaspect
ofthistaskarethecomparablysmallatmosphericamountsexpectedforIO,aroundafewpartsper
trillion(10−12)intermsofvolumemixingratioandprobablyconfinedtothelowestatmospheric
layers.ThesignalsofthesmallspectralIOabsorptionhavetobeseparatedfromnoiseinfluences.
Amultitudeofqualityandconsistencychecksarenecessarytoavoidmisleadingresults.
Withinthepresentstudy,theretrievalofatmosphericIOfromSciamachyhasbeenachieved
(Schönhardtetal.,2007,2008).Thesuccessfulevaluationenablestheobservationofatmospheric
IOcolumnsandtheirspatialandtemporaldistributiononanearlyglobalscale.Observations
becomepossibleinlocationswherenoIOmeasurementshavebeenperformedsofar.Along
termglobaldatasethasbeenanalysed,coveringmorethanfouryearsfromthebeginningof2004
untilmid2008.ThesesatelliteresultsyielddeeperinsightintothepresentIOamountsaswell
aspossibleiodinesources.Severalgeographicalregionsareinvestigated,withonemainfocus
beingtheAntarctic.ConnectionstosimultaneouslyobservedBrOcolumns,theAntarcticseaice
coverageandthephytoplanktonconcentrationsinoceansareanalysed.Thesecomparisonsshall
helptofindlinkstoemissionsources.Forregions,wheretheIOamountsstaybelowthedetection
limit,theidentifiedupperlimitsareusefultoconstrainthepotentialimpactofiodinechemistryin
therespectivelocations.Insomeinvestigations,theexperimentallimitationsarereachedandthe
analysedeffectsremainbelowthedetectioncapability.Currently,Sciamachyistheonlysatellite
forwhichtheretrievalofIOhasbeenmadepossible.OneindependentstudybySaiz-Lopezetal.
(2007a)usesthesameinstrumentandinvestigatestheIOamountsonfourdaysovertheSouth
Region.olarPWithinthescopeofthepresentwork,severalquestionsconnectedwithsatelliteandground-
basedremotesensinghavebeenaddressed.Researchactivitiesincludetheinvolvementinintercom-
parisoncampaignsofground-basedinstruments,dataanalysesandsatellitevalidationactivitiesfor
tracegasesotherthanIO,theplanningofnewinstrumentationscheduledforaircraftmeasurements
ofIOandNO2andrelatedopticaltestmeasurements.Thewrittendoctoralthesisconcentrateson
theretrievalofIOfromSciamachyanddirectlyrelatedresearchaspectsasoutlinedbelow.

thesisthisofOutlineThefirstchaptersummarisesrelevantscientificbackgroundinformation.Thebasiccontextfor
theimportanceofatmosphericiodineresearchisprovidedbydescriptionsofthestructureofthe
atmosphere,theconnectionofhalogenstoozonechemistryandtheroleofatmosphericO3.The
currentstateofatmosphericiodineresearchaswellasprecedingmeasurementsofiodineoxides
aresummarised.Anintroductionintothephysicalprocessesinvolvedintheappliedmeasurement
methodisgiven,andtheutilisedinstrumentsareintroduced.

3

IntroductionandMotivation

Chapter2describesthedevelopmentoftheretrievalofIOfromSciamachydata.Thisincludes
thedataprocessingstepsanddetailsontheDOASretrievalusedfortheIOstandardfit.Crucial
qualityandconsistencychecksfortheretrievalresultsarepresented.Calculationsofthedetection
limitforrelevantsituationsareperformed,andtheuncertaintyonthefinalproductisestimated
fromtheprecisionandaccuracyofthesatelliteIOobservations.Additionallytestedretrievalsnot
fulfillingthequalityand/orconsistencycriteriaarediscussed,ofwhichoneexampleischosenfor
moredetailedanalysisasitexhibitssimilaritiestoanindependentstudy,discussedinChapter4.
Chapter3presentsglobalobservationalresultsoftheIOretrievalfromSciamachy,andthe
mostinterestingregionsarehighlighted.AmainfocusliesonthePolarRegions,inparticular
ontheAntarctic.SouthernHemisphericmapsofthenewlyestablishedIOretrievalexplorethe
temporalandspatialvariationsoverthesea-iceregions,theiceshelvesandtheAntarcticcontinent.
Extractedtimeseriesshowanannuallyrepeatedseasonalcycle.Furtheranalysesconcentrateon
theEastPacifictropicalregionandonNorthernHemisphericcoastlinesathighlatitudes,forwhich
enhancedIOamountsareidentifiedfromthesatelliteobservations.
Chapter4reportsonvalidationandcasestudiesofthesatelliteIOdata.Goodagreementis
demonstratedbetweentheIOobservationsfromthepresentstudyandground-basedmeasurements
atanAntarcticResearchStation.Incontrast,discrepanciesbetweenthepresentstudyandtheonly
otheravailableindependentstudyofIOfromsatelliteovertheAntarcticarediscussed.Datafrom
ground-basedDOASmeasurementsarethencomparedtosatelliteIOresultsforanArcticlocation.
Finally,acarefulselectionprocedureattemptstoidentifythetidalheightdependenceofIOlevels
attwomid-latitudinalsites,whichis,however,belowthedetectionpossibilitiesofcurrentsatellite
ations.observChapter5appliestheavailableCAABA/MECCAchemicalboxmodeltocomputeIOmixing
ratiosfordifferentscenarios.Thequestionisaddressedifpresentmeasurementsofprecursorfluxes
areabletoexplaintheaccomplishedsatelliteresults,andthenecessaryemissionamountsare
determined.Chapter6introducesDOASmeasurementsfromashipcampaignthroughtheAtlantic,where
theinstrumentviewedatanglesbelowthehorizonrecordingthewaterleavingspectralradiance.
Theobjectiveistoimprovesatellitemeasurementsoverwaterbodieswhereproblemshavebeen
identifiedintheretrievalsofseveraltracegases.Acorrectionspectrumisextractedandincluded
insomesatellitetestretrievals.
Asummaryfinallybringstogetherthemainresultsfromthiswork,andanoutlookisgiven
whichproposesfutureactivitiesfortheassessmentofremainingopenquestions.

4

1backgroundScientific

Inthischapter,relevantbackgroundinformationonthesubjectofthisthesisisgiven.Thesci-
entifictopicisintroducedandanoverviewoverthecurrentstateofresearchisestablished.The
majorfocusofthepresentstudyliesoniodinecompoundsandespeciallyonthedetectionofiodine
monoxide(IO).Iodineisrelevantinmanyrespects,andfirstofalltheimportanceforthedifferent
explained.isfieldsTheimportanceofiodineforEarth’satmosphereshallthenbesetinabroadercontext,soan
overviewoftheEarth’satmosphereanditsspecialstructureandcompositionisgiven.Ozone(O3)
isacrucialcomponentinatmosphericcompositionandimportantly,halogensexhibitastrongpo-
tentialforthedestructionofozone.Therefore,therelevanceofO3isexplainedaswellasits
specialroleinthedifferentatmosphericlayersandtheconnectionstohalogenchemistry.Thenthe
mainsourcesofatmospherichalogenspecies,theirchemicalpathwaysandrelevantinfluencesare
describedandthecurrentstateofresearchinatmosphericiodinechemistry.
Thepresentworkincludestherecordingandanalysisofmeasurementdata,forwhichseveralphys-
icalprocessesplayacrucialrole.Opticalmeasurementsofatmospherictracegasabundancesare
performed,wherethebasicdetectionprincipleisthecharacteristicabsorptionoflightbymolecules,
andotherinteractionsoflightwithmatterplayanadditionalrole.Scatteringprocessesandradia-
tivetransferintheatmosphereneedstobeconsideredinordertounderstandthemeasurements.
Thereafter,thespecificscientifictechniqueisdescribedwhichthemeasurementsandanalysesare
baseduponandfinally,theappliedinstrumentsareintroduced.Thereflectionsinthischapteralso
motivatetheactivitiesundertakenwithinthisstudy.

1.1Therelevanceofiodineindifferentfields

Iodineisanaturalchemicalelementandbelongstothefamilyofhalogens.Itwasdetectedinits
elementarystateinthebeginningofthe19thcenturyintheashesofseaweed(Schröteretal.,
1988).Thehalogensmakeuptheseventhmaingroupoftheperiodictableofelements.Iodine
carriesthechemicalsymbolIandanatomicnumberof127(53protons,74neutrons).Theother
halogensarefluorine,chlorine,bromineandastatinewithdifferentimportanceinthevariousfields
ofscience.Althoughelementsofonefamilyoftenhavesimilarpropertiesandundergosimilarchem-
icalreactions,severalcharacteristicsareindividualforeachelement.Especiallywhenconsidering
complexsystemssuchaslivingorganisms,thedifferencesmayshowconsiderableimpactandeach
elementfulfillsauniquerole.Iodineisanessentialelementforvertebrates,hencethesupplyneeds
tobeassured.Thisrelevancealreadyrisesthequestionforsourcesandabundancesofiodineinthe
naturalenvironmentandcausesaconnectionbetweenatmosphericiodineand,e.g.,iodineinthe
humanbody.

5

kgroundbactificScien1

Theappearanceofradioactiveiodinehasaddedimportancetothissubject,asthebiologicaluptake
ofiodineisnotisotopespecific.Radioactiveiodineisbeingreleasedtotheatmosphereinconsider-
ableamountsbynuclearpowerstationsandfromnuclearweaponstests.Throughthefoodchain
andalsothroughtheair,itisdirectlytransportedintolivingorganisms.Itwasradioactiveio-
dineanditsbiologicalthreatwhichoriginallypromotedmeasurementtechniquesforenvironmental
dine.io

1.1.1Iodineanditsrelevanceforhumanandanimalhealth
Consideringthe96naturallyoccurringchemicalelements,25oftheseareknowntobeessentialfor
life.Someofthembuilduptheorganicmaterial,ofwhichcarbon,hydrogen,oxygenandnitrogen
makeup96%,othersarecrucialconstituentsofhormonesorproteins,andsomehavecertainfunc-
tionalityforthenervoussystem.Unlikebromine,e.g.,forwhicharelevantfunctionforvertebrates
isnotknown,iodineisaconstituentoftwocrucialthyroidhormones,triiodothyronine(T3)which
containsthreeiodineatoms,andthyroxine(T4)withfouriodineatomspermolecule.Thethyroid
hormonesplayanimportantroleincontrollingthemetabolicsystemandtheproductionofproteins
andforotherhormonecontrollingfunctions.
Duetotheimportanceofiodinefortheformationofthethyroidhormones,iodineisanessential
elementandthedailynutritionneedstocontaintracesofiodine(about200μgperdayforhumans).
Iodineinsufficienciescanleadtodangerousthyroidhypofunction.Thishealthriskhasbeenidenti-
fiedinthe19thcenturyandiodisedsaltwasproposedtohelpineliminatinghighdegreesofiodine
deficienciesinaffectedcountries(WHO,2007,andreferencestherein).Thenaturalandregular
uptakeofiodinethroughnutritionalsopermitsradioactiveiodineisotopestoentertheorganismin
casetheyareavailable.

dineioRadioactive1.1.2Severalradioactiveisotopesofiodineexist,e.g,iodine-129andiodine-131.Theseisotopesare
emittedbyhumanactivities,mainlyfromnuclearweaponstests,accidentsorleakagesinnuclear
powerstationsandfromnuclearfuelreprocessingplants.Thehalf-livesvaryquitestronglywith
15.7millionyearsforiodine-129and8daysforiodine-131.Withitscomparablyshorthalf-life,
iodine-131hasadditionalmedicalrelevanceintherapiestospecificallycurethyroidhyperfunction
(overactiveproductionofT3andT4)butalsoasadiagnostictracer.
Theavailabilityofradioactiveiodineafternuclearaccidentsposesathreattoanimalhealth,as
itisingested(orinhaled)justinthesamewayasstableiodine-127andcanaccumulateinthe
thyroids(RobertsonandFalconer,1959).Followingtheatmosphericnuclearweaponstestsinthe
1950sand60s,largeamountsofradioactiveiodineenteredtheatmosphere(Chamberlain,1960).
Thepresenceofthisencouragedresearchonthefieldofiodinechemistry,especiallyitspathwaysin
theenvironment(Chamberlainetal.,1960)havetobewellknowninordertoestimatethepotential
healthriskarisingfromradioactiveiodine.Throughextensivebiological,chemicalandgeological
cycles,thesespeciesareofrelevancealsoatfurtherdistancefromtheliberationsite.Bothspecies,
iodine-131andiodine-129,canundergomanychemicalreactionsandentervariousdomainsbetween
theatmosphereandthebiospherebeforetheyhavetransformedtothestable127-isotope.

6

1.1.3Therelevanceofiodineforthebiosphere

1.2IntroductiontoEarth’satmosphere

Withinplantsandalgaeandpossiblyalsoforanimals,iodineandiodinecompoundshavebeen
foundtoprotectthelivingorganismfromoxidativedecay.Reactiveoxygenspecies(ROS),which
arepartlyconvertedtohydrogenperoxide,causeoxidativedamagetoorganiccells.Iodideions(I−)
havetheabilitytoscavengeROSthuspreventingcelldecay(Küpperetal.,1998).Inthereaction
ofiodidewithhydrogenperoxide,iodinatedformsoforganicsubstancesareproduced,whichare
harmlessfortheorganisms.Especiallyorganismslikebrownalgaeaccumulateiodinespeciesin
theircellsanddrivetheenvironmentalcyclingofiodinethroughtheemissionoforganiciodine
speciessuchasiodomethane.Recently,ithasbeendiscoveredthattheaccumulatedforminthe
algaespeciesisactuallyiodide(Küpperetal.,2008).
Duetothefactthatiodinespeciesshowsuchanimportanceforthebiosphere,anduptakefrom
thegaseous,liquidandsolidphasesaswellastheemissionofiodinecompoundsbackintothe
atmosphereorhydrospheretakesplace,onereferstothebiogeochemicalcyclingofiodineinthis
respect.Bothdirections,thepathwaysfromtheatmospheretothebiosphereandviceversaplay
animportantrole.

1.1.4Therelevanceofiodineintheatmosphere

Adetailedoverviewofthecurrentstateofresearchinthefieldofatmosphericiodinechemistryis
subjectofSection1.5.Inshort,iodinehasanimportantinfluenceintwoaspects:

•Throughthereactionwithozone,iodinealtersthechemicalcompositionoftheatmosphere
anditsoxidizingcapacity.

•Iodineoxidesleadtotheproductionoffineparticleswhichmayinfluencetheradiationbudget.

Mostimportantly,ozonemoleculesaredestroyedinthereactionwithiodinewherebyiodinemonox-
ideisformed.Althoughoverallamountsofiodinearerathersmall,itsrelevanceisenhancedby
catalyticozonedestructioncycles(cp.Sec.1.5.2).Aftersomefundamentalpropertiesoftheatmo-
spherearedescribedinthenextsection,theconsiderationswhyozoneisimportantintheatmosphere
1.3.Sec.insummarisedare

1.2IntroductiontoEarth’satmosphere

Theatmosphereofaplanetistheshellaroundtheplanet’smainbodycontainingamixtureofmainly
gaseoussubstances.Whilenoteveryknownplanetexhibitsanatmosphere,theatmospheresare
uniquefortherespectiveplanet.TheatmosphereoftheEarthexhibitsanexceptionalcomposition
whencomparedwiththeatmospheresofotherplanetsintheSolarSystem.Basicinformationon
thestructureandcompositionoftheatmospherecanbefoundinstandardtextbooks,e.g.in
(2000).yneaW

7

kgroundbactificScien1

ositionCompatmospheretheof

ThespecialcompositionofEarth’satmosphereisaconsequenceoftheexistenceoflifeonEarth.
Thecurrentstateoftheatmosphereintermsofthermodynamicsisasteady-statedisequilibrium
anditishighlyreactive.Thisstateiscausedbythebiologicalprocesses,whichareresponsiblefor
thehighamountofoxygenintheatmosphere.Withoutlife,theamountofoxygenwouldbe1013
timessmallerthanpresentlyobserved.Oxygenisareactivespecies,sothatmanypossiblechemical
reactionstakeplace,inwhichtheamountofoxygenisreduced,butthelivingorganismskeepupthe
highoxygenconcentrationscontinuously.Withoutlife,allpossiblechemicalreactionswouldtake
placeuntilachievementofchemicalequilibrium.Thedryatmosphereconsiststo99.9%ofthemain
speciesnitrogen,oxygenandargon(O2,N2andAr),whiletracespeciesmakeuponly0.1%ofthe
dryatmosphere.Table1.1liststhecurrentcompositionofEarth’satmosphere(Wayne,2000).The
amountsaregivenintermsofvolumemixingratio(VMR),i.e.thevolumeofspeciesconsideredin
relationtothetotalvolumeofair.Forsmallamounts,theshortnotationofpartspermillion(i.e.
10−6,ppm),partsperbillion(i.e.10−9,ppb)orpartspertrillion(i.e.10−12,ppt)iscommonlyused.

Inadditiontothegaseousspecies,theatmospherecontainsliquidorsolidsubstancessuspended
inair,whicharesummarisedasaerosols.Aerosolsmayhavevariouscomposition,sizeandshape.
Ontheirsurfaces,certainchemicaltransformationsareinduced.Mostimportantly,theyinfluence
theradiationbudget,whichisaresultoftheirlightreflectingandinsomecaseslightabsorbing
properties.Overall,thecurrentknowledgeisthataerosolscauseanegativeclimateforcing,i.e.,
onaveragetheyinduceacoolingeffect(IPCC,2007).Thefactthataerosolscanactascloud
condensationnucleiintensifiestheirimportanceforEarth’sclimatestate.Thecompositionandthe
amountofaerosolsishighlyvariablewithtimeandspaceandtheirexactclimateimpactisyetto
determined.eb

effectgreenhouseThe

WhiletheequilibriumtemperatureoftheEarth’ssurfacewouldbearound255K(-18◦C)without
theexistenceofthegreenhouseeffect,certaingasesintheatmospheresuchaswatervapourcausea
naturalgreenhouseeffectof+33K,givinganaverageglobalsurfacetemperatureof288K(+15◦C).
Thegreenhouseeffectisbasedonthesolarandterrestrialradiationproperties.Thesunemitsa
modifiedblackbodyspectrumwithaneffectivetemperatureofabout5780Kandaspectralmax-
imuminthevisiblewavelengthrange(cp.Sec.1.6).Earthisaninfraredemitteratitseffective
radiationtemperatureof255Kandaspectralmaximumatapproximately10μm.Substanceswhich
don’taffecttheincomingsolarradiationmuchbutabsorbenergyintheterrestrialinfraredspectral
regioncontributetothegreenhouseeffect.Themostimportantnaturalgreenhousegasesarewater
vapour,carbondioxideandmethane.Someothergreenhousegaseshaveamuchlargerwarming
potentialpermoleculebutarejustnotasabundant.ThereflectingpropertyoftheEarth’ssurface
butalsoofaerosolshasanadditionalinfluenceonthegreenhouseeffect.Astheanthropogenic
impactonclimatehasbecomeapparentoverthelastdecadesandprocesseslikeglobalwarmingare
takingplace,largeresearcheffortsaremadetobetterquantifythedifferentinfluencingfactorsand

8

1.2IntroductiontoEarth’satmosphere

VMReciesspGaseciesspMainNitrogen(N2)78.08%
Oxygen(O2)20.95%
0.93%(Ar)ArgonTracespecies,temporallyandspatiallyconstant
NeonHelium(Ne)(He)185.2ppmppm
ppm1.0(Kr)KryptonHydrogen(H2)0.5ppm
Tracespecieswithvariableamounts,averagevalues
Watervapour0-4%
Carbondioxide(CO2)379ppm(IPCC,2007)
Methane(CH4)1.7ppm

T(2000)ableand1.1:OvIPCCerview(2007)overarethegivencompinositionpartspofercenEarth’storperatmosphere.millionbyNumvbolume.ersasreportedbyWayne

terrelations.intheirThecentralspeciesofthepresentstudy,iodinecompounds,donotactasgreenhousegasesthem-
selves.However,thesecondaryeffectofparticleformationhasapotentialinfluenceontheclimate
state(O’Dowdetal.,2002b).

Thelayeredstructureoftheatmosphere
Thegeneralstructureoftheatmosphereisrelevantfortheunderstandingofthespatialdistribution,
especiallytheverticallocation,ofprocessesandsubstances.Theverticalstructureoftheatmosphere
canbedescribedbydividingitintosubsequentaltitudelayers,eachwithindividualcharacteristics.
Atypicaltemperatureprofileofthemid-latitudeatmosphere(U.S.standardatmosphere)is
illustratedbyFig.1.1,showingthedifferentlayerswhicharedeterminedbythetemperaturestruc-
ture.Inaddition,thealtituderangeofthelargestozoneconcentrationisindicated.
Themainpartofterrestriallifeisconcentratedinthelowestlayersoftheatmosphere.In
thetroposphere,whichreachesfromthegrounduptothetropopause,thetemperaturetypically
decreaseswithaltitude.Strongmixingprocessescharacterisethetroposphereaswellasdirect
interactionswiththeothercomponentsoftheEarthsystem(geosphere,hydrosphere,cryosphere
andbiosphere).Thisisespeciallyvalidfortheso-calledboundarylayer,whichisthesublayerclosest
totheEarth’ssurface,wherefrictionfromthesurfacehasamajorinfluenceonthedynamicsin
contrasttothefreetroposphereabovewherefrictioncanbeneglected.Theboundarylayer(BL)
thereforestandsindirectcontacttothesurface,theoceansandtoplantsandanimalsandits
compositionisimmediatelyaffectedbyhumanactivities.Intermsofchemicalcomposition,the
boundarylayeroftenexhibitsindividualproperties.Severalrecentresearchstudiessuggest,that
iodinechemistryismainlylocatedintheboundarylayerandhasnegligibleinfluenceinhigher
altitudes.Thealtitudestructurevarieswithlatitude,seasonandotherparameters.Whilethetroposphere

9

kgroundbactificScien1

TTherhermomospherspheree
MesopMesoppaauussee

MeMesososphsphereree
SSttrraattoopppaauussee
SSttrraattoospspherheree
TTrropopppopopauausese
TTrrooppoossphphereree

ypicalT1.1:Figureofprofileeraturetempatmo-mid-latitudetheU.S.(standardspherehwhicatmosphere),tdifferenthedetermineslayers.Thetemperature
influ-cruciallyisprofilemmaax. Ox. O33ccooncenncentrtraattioionnencedbytheozonelayer
stratosphere.theinBoundaryBoundaryyLaLayyyeerr

hasatypicalheightof10kminthemid-latitudes,itisloweratthepoles(around8km)andhigher
attheequatorwithupto15km,duetolargescaledynamics.Theheightoftheboundarylayer
isalsohighlyvariable.Inthemid-latitudesvaluesontheorderof1km,varyingbetween500m
and2km,arecommon.ThestructureoftheatmosphereintheAntarcticisspecialinseveral
aspects.Descendingmotionsoverthepolescompressthelayerscausingalowertropopause.Also
theboundarylayerisshallowerwithausualthicknessaround200m.
Atthetropopause,thetemperaturedecreasefromthetropospherebelowceasesandisslowly
revertedintoatemperatureincreaseinthestratosphereabove,causedbythestrongabsorptionof
solarUVradiationbyozoneinthestratosphere.Thestratosphereisthereforecomparablystable
andstrongmixingislargelyprohibited.Nevertheless,somemixingprocessesandexchangebetween
theuppertroposphereandlowerstratosphere(UT/LSregion)takeplace.Inthestratosphere,the
ozonemixingratioexhibitsitsmaximumvaluesformingtheozonelayer,whichiscrucialformost
1.3).Sec.(cp.organismslivingThepresentstudymainlyfocusesonprocesseslocatedinthetroposphere.Insomeaspects,the
higheraltitudesbecomeimportantnevertheless.Thechemicalfamilyofthehalogens(especially
chlorine,bromineandiodine)areofrelevanceindifferentaltitudelayers.Theimportanceofhalogen
speciesforthechemicalcompositionhasbeenrecognizedespeciallyinconnectionwithozone.The
mostimportantaspectsofatmosphericozoneisaddressedbythefollowingsections.

1.3Theimportanceofozone
Ozone(O3)isanimportantchemicalintheatmosphere,especiallywithregardtolifeonEarth.
ThemostrelevantpropertyofO3isitsstrongabsorptionofradiationintheultraviolet(UV)
spectralrange.ForUVradiationbelow230nm,theabsorptionbyoxygenisstrongenoughtoavoid
penetrationofthispartofthesolarspectrumdowntotheEarth’ssurface.Thecrucialwavelength
regionliesbetween230and290nm,wherebiologicallyimportantmoleculeswouldstillexperience
severedamageandtheprotectionbyoxygenisnotstrongenough.Ozonehappenstobetheonly

10

1.3Theimportanceofozone

chemicalintheatmospherewithrelevantabsorptioninthisspectralregion.O3thereforeactsasa
veryimportantUVshield.Duetoitsmainformationanddestructionpathways,O3showsaspecial
altitudeprofilewithadistinctmaximuminthestratosphere.
SeveralpiecesofinformationinthisandthefollowingsectioncanbefoundinWayne(2000)and
Brasseuretal.(1999),othersourcesarecitedindividually.

theinOzonestratosphereThefirstschemeofO3formationanddestructionwasproposedbyChapman(1930)andwasex-
tendedlaterduetosomemissingO3lossmechanisms.TheChapmanschemeismainlygivenby
fourreactions,thefirstonecreatingoddoxygen(OorO3),thelastonereducingoddoxygenand
twowithnochangeinOandO3.Hereandinotherfollowingreactions,Misathirdbodyfor
energyandmomentumtransferduringcollision,typicallyN2orO2molecules.

O2+hν→O+O(R1)
O+O2+M→O3+M(R2)
O3+hν→O+O2(R3)
O+O3→2O2(R4)
Reaction(R1)representsaphotolysisreaction,wherehνistheenergyoftheincidentphotonwith
Planck’sconstanth=6.626·10−34Jsandfrequencyν.Theratewithwhichaphotolysisreaction
takesplaceisdeterminedbytheconcentrationofthegastobephotolysedandthephotolysisfre-
quency,e.g.JR1,whichitselfiscalculatedfromtheabsorptioncrosssection,thequantumyieldand
radiation.incomingtheAs(R4)wasfoundtobetooslowtoeffectthenecessarydecompositionofozoneinordertobalance
O3productionandtoachieveagreementwithobservations,additionalmechanismswereproposed.
Basically,theyfollowasimplecatalyticcycle,withthespeciesAfacilitatingO3decomposition:

A+O3→AO+O2(R5)
AO+O→O2+A(R6)
Net:O3+O→2O2
Thenetreactiongivesthesameresultas(R4),butismediatedbycatalystA.ThesubstanceA
doesnotneedtobepresentinlargeamountstobeeffective,becauseitisnotconsumedinthecycle
andmayreactagaininthesamescheme.Specieswhichcanactintheabovereactionsascatalyst
Aincludeatomichydrogen,thehydroxylradical,nitrogenmonoxide,aswellashalogenatoms(H,
OH,NOaswellasCl,BrandI).Dependingonthealtitude,thesespecieshavedifferentimportance
fortheO3lossrate.Whilechlorineismostinfluentialinthestratosphereandbromineissignificant
inboth,thestratosphericandtroposphericlayers,iodinechemistryprobablytakesplacemainlyin
thelowestpartsofthetroposphere.Theoverallchemicalschemeforozoneisfurthercomplicated
duetoreactionsbetweenthedifferentcatalystfamilies.
WhilesomeofthecatalystsexistnaturallyanddeterminetheoriginalamountsofO3,especially
theabundancesofhalogenspecieshaveincreasedduetohumanactivity.Thenatural,mainly

11

tificScien1kgroundbac

oceanicsourcesofhalogensarebyfarsmallerthanmanmadeemissionsfrombiomassburning
andespeciallyindustrialactivities.Especiallytheuseofchlorofluorocarbons(CFCs)increased
atmosphericburdensofchlorinebeforetheMontrealprotocolin1987reducedemissionsofCFCs.
However,largeamountsofCFCshavealreadybeenemittedtotheatmosphereandareeventually
transportedtothestratosphere.Whiletheyareextremelystableinthetroposphere,CFCsaree.g.
photolyticallydecomposedinthestratosphereandreleasereactivechlorineatomsforcingtherapid
cycles.destructionozoneEspeciallyovertheAntarcticinspringtime,ozonecolumns(i.e.,totalozoneamountsintegrated
fromthesurfacetothetopoftheatmosphere)nowadaysdramaticallyreducebytypicallyafactor
ofthree.OriginalvaluesforAntarcticspringtimeO3columnswereintherangeofmorethan300
Dobsonunits(1DU=2.65×1016molec/cm2,i.e.,a10μmhighcolumnatstandardtemperature
andpressure),whileuptotheendofthe20thcentury,springtimeO3columnsreducedtoaslow
asaround100DUinsomeyears.TheconcentrationofO3evendropstozeroincertainaltitudes.
Situationswithcolumnamountslowerthan220DUarereferredtoasozoneholes.
Thecompletemechanismofozonedestructionhasbeenrevisedmanytimes,asmoreknowl-
edgeonthereactionpathways,concentrationsofspeciesandmorepreciselaboratorydatabecame
available.Uptotoday,theexactschemeisnotfullyclarified.Themainreactioncyclesdestroying
ozoneinthestratosphereinvolveClatoms,andadditionallythespeciesOHandHO2(HOx)as
wellasNOandNO2(NOx)withcombinedcatalyticreactioncycles.Furthermore,brominecycles
involvingBratoms,e.g.inducedbybrominatedCFCs(halons)mediateozonedestruction.Levels
ofbrominecompoundsarebyfarlowerthanthoseofchlorinespecies,butthepotentialofozone
depletionbybromineisverylargeandthepresenceofBratomsevenenhancestheeffectofchlorine
onozonedestructionbycrossreactions.
Thestrongozonelossseeninozoneholesituationsrequirescertainsurroundingconditionsin
additiontothepresenceofozonedepletingatomsandmolecules.Theseconditionsaremainly
presentinAntarcticspringtimeandincludeverylowtemperatures,theformationofpolarstrato-
sphericclouds(PSCs)andthestablepolarvortex,anearlyenclosedregionformedbylargescale
dynamicswhichkeepsuptheseconditionsforperiodsofseveralweekstomonths.Itisthesurfaces
ofthePSCs,whichprovideconditionsforadditionalheterogeneousreactionsstronglyenhancing
theozonelossbyconvertingreservoirspecies,i.e.fairlystable,non-reactivecompoundsbackinto
theactivecatalysts,theClatoms(cp.Sec.1.4.1).
Theozonelossinthestratospherevariesfromyeartoyeardependingonthemeteorological
anddynamicconditions.Theconcentrationsofcatalyticallyactivespecieswillprobablyfurther
decreaseinthefuture,butthetimescaleisratherlargeduetothelonglifetimesoftheprecursor
substancesbeforereachingthehighaltitudes.

ospheretroptheinOzoneInthetroposphere,ozonemixingratiosarealotsmallerthaninthestratosphericozonelayer.The
importanceoftroposphericozonehastwoaspects.Ononehand,ozoneposesaseverehealthrisk
tohumans,animalsandplantsasitbecomespoisonousforlivingcellsabovecertainlimits.Onthe
otherhand,however,ozoneintheloweratmosphericlayersisneededasamainproducerofOH

12

atmospheretheinHalogens1.4

molecules.Below310nm,O3photolysestoproduceexcitedoxygenatoms,O*(1D),whichreact
withwatervapourtoyieldOH:

O3+hν→O2+O∗(1D)(R7)
O∗(1D)+H2O→OH+OH(R8)
OHisaveryinfluentialmoleculeinatmosphericchemistry.Itisextremelyreactiveandcapable
ofinducingefficientoxidationprocesses.Inthisrespect,OHisthemostefficientagentfordecom-
posingairpollutionchemicalsasitoxidisesmostchemicalsfoundintheatmosphere.OHoxidises
COandCH4,leadingtotheproductionofperoxyradicalsandsubsequentlyperoxides.AlsoSO2
andNO2areremovedfromtheatmospherebyreactionwithOH,formingsulphuricandnitricacid,
whicharewashedoutbywetdeposition.
TheamountofO3inthetroposphereisinfirstplacedeterminedbytheNOxratioduetothe
cycle:reactionequilibriumwingfollo

NO2+hν→NO+O∗(1D)(R9)
O∗(1D)+O2+M→O3+M(R10)
O3+NO→O2+NO2(R11)
FromtheNOxratio,thephotolysisfrequencyJR9andthereactionratecoefficientkR11,the
equilibriumO3concentrationmaybecalculated.However,theaboveequilibriumisdisturbedby
thepresenceofadditionalchemicalsubstances,whichleadtoeitherO3productionordestructionin
thetroposphere.Brominechemistryandalsoiodinechemistry,e.g.,reducetroposphericozonecon-
centrations(Dickersonetal.,1999;Readetal.,2008).Ontheotherhand,ozonemaybeeffectively
producedinthepresenceofNO2andperoxides(RO2,withe.g.R=H,R=CH3).Thephotolysis
ofNO2providesthenecessaryoxygenatomsasabove,whileRO2reactionschemesconvertNO
backtoNO2.InpollutedregionsandwithanincreasedburdenofRO2fromorganicprecursors,
theadditionalozoneproductionleadstothephenomenonofsummersmog.TheO3concentration
limitsintheEuropeanUnionare180μg/m3and240μg/m3forinformationandwarningofthe
population,respectively(EuropeanParliament,2002).

Inconclusion,stratosphericozoneisessentialforlifeonEarth,whileinthetroposphereitsroleis
biguous.ammore

atmospheretheinHalogens1.4

Fromthehalogenfamily,mainlyCl,BrandIplayimportantrolesforatmosphericchemistry,
whilefluorineformsverystablereservoirspecies(especiallyHF),andastatinehasextremelylow
abundances.Thethreerelevanthalogensshowpartiallysimilarreactionsandinfluences,butalso
exhibitsomeindividualproperties.Chlorineandbrominewereearlierfoundtohaveatmospheric
relevancethaniodine,andsomekeypropertiesofthesetwospeciesshallbediscussedfirst.Due
toanalogiesandinteractionsbetweenthedifferenthalogentypesalsochlorineandbromineare
relevantwhenanalysingtheroleofiodineintheatmosphere.Atmosphericiodinechemistrywillbe

13

tificScien1kgroundbac

addressedindividuallyandindetailinthenextsection.
Itwasinpartthedetectedinfluenceofhalogensonstratosphericchemistrythatdroveresearchin
tropospherichalogenchemistry.Tounderstandtheatmosphericrelevanceofasubstanceandto
estimatepossiblefuturechanges,itssourceshavetobeknown.Fromemittedcompoundsinthe
troposphere,long-livedspeciesmaybetransportedintothestratosphere.

1.4.1Halogensinthestratosphere
Themaininputofhalogenstothestratosphereresultsfromtransportoflong-livedCFCsorhalons
upwardsfromthetropospherewheretheywereliberated.Additionally,short-livedcompounds(or
veryshort-livedsubstances,VSLS,severalbrominatedandchlorinatedcarbons)wereidentifiedto
contributetothestratospherichalogenload.Thehalogencompoundsarephotolysedinthehigh
altitudesofthestratosphereandreleaseClandBratoms.Alternatively,chemicalbreak-upofCFCs
takesplace,e.g,inthereactionwiththeO(1D)radical(Ravishankaraetal.,1993),fromwhichClO
maybereleased,therebycontributingtothereactivechlorinebudget(Brasseuretal.,1999).As
soonasthebreak-uphasstarted,acomplexcatalyticchemicalreactionmechanismisactivated.
Themajoreffectofthismechanismandhenceofhalogensinthestratosphereisthedestructionof
ozone:stratosphericCl+O3→ClO+O2(R12)
SubsequentrecoveryoftheClatom(equivalentlyforBr)viadifferentpathwaysleadstoacatalytic
cycle.OneimportantpathwayincludesreactionsonthesurfacesofPSCparticles,shortlymen-
tionedinSec.1.3.PSCparticlesconsisteithermainlyofHNO3(typeI)orofH2O-iceandsome
HNO3-hydrates(typeII).Inbothcases,afrozenorliquidaerosolcoremaybepossible.Afterre-
actionofClOwithNO2,chlorinenitrateforms(ClONO2)whichcanreactwithHCl(fromvolcanic
eruptionsorchemicalconversions)ontheparticlesurfaces:

ClONO2+HCl→Cl2+HNO3(R13)
Cl2+hν→2Cl(R14)
Thisisanimportantreaction,astworeservoirspecies(ClONO2andHCl)areconvertedatthe
sametimeandtworeactivechlorineatomsarereleasedthatcanenter(R12)again.Additional
reactionswhichoccuronPSCsurfacescreateacomplexmechanismwhichisnotyetcompletely
understood.Stillsomeratecoefficientsandreactionpathwaysaredebated.
Thepresenceofbromineleadstoozonedepletionintwoways.Bratomsdirectlydestroyozone,and
additionallyanintensificationofthechlorinecyclesthroughreactionsofBrOwithClOiseffected:

BrO+ClO→Br+Cl+O2(R15)
Whilealsodifferentproductpathwaysarepossible,thisexampleshowshowthecrossreactionleads
tonewreleaseofhalogenradicals.WhencomparingthenumberofO3atomsonaveragedestroyed
perhalogenatom,bromineismoreeffectivethanchlorineapproximatelybyafactorof50(Wayne,
2000).Fromtheirmaximumabundanceinthe1990s(forsomecompoundsmuchlaterinthe2000s),

14

atmospheretheinHalogens1.4

theamountsofseveralstratospherichalogenspecieshavestartedtodecreaseslightly(WMO,2006).
Presently,theamountsofbrominatedspeciesinthestratosphereliearound18to25ppt,ofwhich
5pptresultfrombrominatedVSLS.Amountsoftotalavailablechlorineinthestratosphereare
3.5ppbwiththelargestamountfromthelong-livedCFCsandsmallercontributionfromchlorinated
VSLS(around50ppt)(WMO,2006).Asdiscussedlater(cp.Sec.1.5),therelevanceofiodinein
stratosphericchemistryismostprobablysmall,maybeevennegligible,butremainstosomeextent
uncertain.

1.4.2Halogensinthetroposphere

Severalsourcesofhalogenspeciesorhalogenatedcompoundsareknown.Afterprimaryrelease,
chemicalconversionsleadtoformationoffurthersubstances.Halogensingeneralareboth,of
naturalandanthropogenicorigin.

sourcesNatural

Oneoriginalsourceofnaturalhalogencompoundsaretheoceans.Seasaltisrichespeciallyin
chloride(Cl−),butalsocontainsbromideandiodide(Br−andI−).Thenumberdensitiesarequite
differentthough,asconvertedfromWayne(2000):

[Cl−][Br−]
[Br−]≈660,[I−]≈15000.
Seasaltaerosolscontainlessbromideandchloridethanexpectedfromtherespectivesodiumcon-
tent,sodirectinorganicreleaseofhalogencompoundsfromseasaltparticlesseemsprobable.Via
photolysis,thesemoleculesyieldreactivehalogenatoms.Othergenerallyhaliderichdomainsare
soils,inlandsaltwaterlakesandsaltflats.

Coastalareasandtheopenoceanaresourcesofseveralvolatilehalogenatedorganiccompounds-
suchasCH3Cl,CH3BrandCH3I,forexample.Alsopolyhalomethanes(e.g.,CH2Br2,CH2I2,
CH2BrI,CHBr2Cl,etc.)arereleasedbybiologicalprocessesinoceans(ReifenhäuserandHeuman,
1992;Carpenteretal.,1999).WhileCH3ClandCH3Brhaverelativelylonglifetimesinthe
troposphereandcanbethereforetransportedtothestratosphere,CH3Iismoreeasilyphotolysed
withatypicallifetimeofseveraldaysandreleasesIatomsmostlyinthetroposphere.CH3Clisthe
largestchlorinesourcegasingeneralandoriginatesto10%fromtheoceansandto80%frombiomass
burning.Theremaining10%isproducedbyindustrialactivities(Wayne,2000).Biomassburning
mayalsoproducesomeCH3Iamounts(Andreaeetal.,1996).Volcaniceruptionsareavariable
sourceofhalogenatedspecies.Thetiming,strengthanddurationoferuptionsanddegassingperiods
arefluctuating,andalsothehalogencontentintheoutburstsanddegassingprocesseschangeand
differsbetweenindividualvolcanos.Emissionsfromvolcanoescontainlargeamountsofhydrochloric
acid,HCl,aswellasotherchlorineandbrominespecies(Francisetal.,1998;Bobrowskietal.,2003).

15

kgroundbactificScien1

sourcesogenicAnthropApartfromthenaturalsources,thereisasubstantialanthropogenicinfluenceonatmospherichalo-
genlevels.Methylchloridefromboth,naturalandanthropogenicsources,ispresentatanaverage
mixingratioof0.5ppb.TheaforementionedCFCs,whichareusedforindustrialpurposes,have
increasedtheburdenofhalogenspeciesdramatically.Industrialapplicationsincludetheuseas
refrigerants,solventsorfertilizers.FamousexamplesofCFCsareCFCl3andCF2Cl2,whichcarry
tradenamessuchasFreon-11andFreon-12.EnormousamountsofCFCswereemittedtotheat-
mospherebeforethemid1990s,andduetotheirchemicalstabilityinthetroposphere(withlifetimes
longerthanhundredyears),themoleculesarenotalteredbeforetheyareeventuallytransportedup
tothestratosphereandphotolyticallyreleaseClatoms.
Consideringbromine,animportantanthropogenicsourcearethehalons,i.e.brominatedCFCs.
Thesecompoundshavebeenusedasfireextinguishers,withCF2BrClandCF3Brbeingthemost
commonsubstancesofthisfamily.Theirtroposphericmixingratiosamounttoseveralppt.Atmo-
sphericmethylbromide(CH3Br)is,apartfromitsnaturalsources,producedbyhumanactivities,
e.g.byuseinagricultureandbybiomassburning.CurrentCH3Brmixingratiosliearound10ppt.

Somerelevantpathways
Severalindustriallyproducedchlorineandbrominespeciesaretoahighdegreechemicallyinertin
thetropospheresothattheydonotinterferewithtroposphericcomposition.Othershowever,are
tosomeextentphotolabilealsoatwavelengthsthatreachdowntotheEarth’ssurface.CH3Cland
CH3Br,forexample,releaseClandBratomsalsointhetroposphere.
Halogensareefficientoxidants.Oneimportantexampleofoxidationpathwaysistheoxidationof
gaseousmercury(Hg)by,bromineatoms,whichhasbeenobservedinpolarregions.Theoxidated
formofmercuryismoreeasilytransferredtosnowandothersurfacesandcanbeincorporated
bybiologicalorganisms.Mercuryispoisonousalreadyinverysmallamounts,whichmakesthis
transformationadangerousprocessthreateningthebiosphereintheaffectedregions.
Inaddition,halogensreactwithorganiccompounds,e.g.,hydrogenatomsarereplacedby
halogensinorganiccarbons,formingthestartofoxidationchains.
HalogenatomsreactwithO3leadingtoozonelossalsointhetroposphere.Theimportance
ofozoneforthetropospherehasbeendiscussedabove,andthedestructionofO3hasastrong
influenceonthechemicalcomposition.Inthemid1980s,theconnectionbetweenstrongozone
depletionevents(ODEs)inPolarRegionsandthepresenceofbrominecompounds(inthatcaseof
filterablebromine)hasbeendetectedatBarrowintheArctic(Barrieetal.,1988).Theseevents
wereobservedinPolarSpring.Inmanycases,theozonemixingratioshowsananti-correlationwith
measuredbromineamounts.Thereleaseandoverallprocessoftheseeventshavenotbeencompletely
resolvedyet,butseveralmechanismshavebeenproposed.Mostlyinorganicreleaseprocessesfrom
seaicecoveredregionsareconsidered,eitherfromsea-icesurfacesdirectly,orfromfrostflowers,
aerosolsorbrine(Kaleschkeetal.,2004;Sanderetal.,2006a;Simpsonetal.,2007a;Piotandvon
Glasow,2008).Possibly,thelowtemperaturesoftheaerosolsorsurfacesintherespectiveregions
areimportantforthereleasemechanism.Asthelocalamountsofbrominecompoundsoftenincrease

16

1.5Currentstateofatmosphericiodineresearch

veryfast,thefollowingmechanismhasbeenproposedasapossibleexplanationfortheobservations.

Bromineexplosionandozonedepletionevents
Theso-called”bromineexplosion”isanautocatalyticandmulti-phase(m.p.)reactioncyclecausing
aquickandefficientreleaseofbromineatomstotheatmosphere,moreorlessdirectlyfromthesea
saltcontent(PlattandHönninger,2003;Simpsonetal.,2007a):

HOBr+Br−+H+m.p.→H2O+Br2(R16)
Br2+hν→2Br(R17)
Br+O3→BrO+O2(R18)
BrO+HO2→HOBr+O2(R19)
Net:H++Br−+HO2+O3→Br+H2O+2O2
TheabovereactionschemethenleadstoexponentialincreaseintheconcentrationofBrO,whichis
oftenpresentatlevelsofseveralppt(Simpsonetal.,2007a).Nobiologicalprocessesareinvolved
inthisproposedmechanism.Oneimportantprerequisiteforthiscycletohappenissomesufficient
acidity(involvementofH+ions).Thecrucialreactioninthisschemeisthefirstmulti-phasereaction
astwoinactivebrominespeciesareconvertedwithinseawaterorseasaltaerosoltoBr2,which
enterstheatmosphereandpotentiallyyieldstwohighlyreactiveBratomsafterphotolysis.Satellite
observationsofBrOshowthatoftenlargeareasexhibitenhancedBrOamounts(WagnerandPlatt,
1998;Richteretal.,1998),andtheseareasareconnectedtoregionscoveredwithseaice(Kaleschke
etal.,2004).ThisprocessoccursinasimilarwayonbothHemispheres,intheArcticandthe
Antarctic,beginningwithPolarSunriseandlastingforsomemonths.

1.5Currentstateofatmosphericiodineresearch
Althoughiodinespeciesaretypicallylessabundantthanchlorineandbrominespecies,theyhave
receivedincreasingattentionduringthelastyears.Thisisinpartdrivenbytheawareness,thatlarge
effectsmayariseevenfromsmallabundances,e.g.,throughcatalyticcyclesorbyiodinespecific
pathways.Thebiologicalimportanceofiodineandthepresenceofradioactiveiodinehasalways
addedtotheneedofunderstandingiodinerelatedprocesses.
Insomeoccasions,troposphericiodinecompoundshavebeenobservedatsimilarlevelsas
equivalentbrominecompounds,whichhintsatveryefficientreleasemechanismsoreveniodine
specificpathways.Additionally,throughcrossreactionsbetweeniodineandbrominecompounds
whichmayleadtothere-releaseofreactivebromineatoms,thepotentialinfluenceofbromineis
increased.Inthepastyears,considerableprogresshasbeenmadeonthefieldofiodineresearchandin
understandingtherelevanceofhalogens,especiallyinthetroposphere,whilemanyopenquestions
stillremainandaskforfurtherresearchefforts(PlattandvonGlasow,2005).Inthefollowing,
centralaspectsofatmosphericiodinechemistry,someidentifiedsourcesandprecursors,aswellas
theprocessofparticleformationandobservationsofiodineoxidesaresummarised.

17

1kgroundbactificScien

1.5.1Sourcesofreactiveiodinecompoundsintheatmosphere
Whileiodineisasolidsubstanceatusualatmospherictemperatures,atomicandmoleculariodine
(IandI2)existingaseousformafterreleasefromspecificprecursors.Atomiciodineandiodine
oxidesareradicalsandconsequentlyreactivegases.Onemaindiscussionisconcernedwiththe
quantificationoforganicsourcesononehandandinorganicpathwaysontheother.Althoughnot
allquestionshavebeenansweredyet,largeresearcheffortshaveledtoanincreasedknowledgeand
understandingoverthepastdecades.
Alreadyinthe1970s,atmosphericabundancesofmethyliodide(CH3I)wereobservedbyLove-
locketal.(1973)inandovertheAtlantic.MeasurementactivitieshavesincefoundCH3Iinsev-
eralotherlocations,aswellasnumerousadditionalorganoiodinecompounds,e.g.,diiodomethane
(CH2I2),iodochloromethane(CH2ClI),andpropyliodide(C3H7I)(Rasmussenetal.,1982;Reifen-
häuserandHeuman,1992;Schalletal.,1994).Thesesubstancesaresummarisedasvolatileorganic
iodine(VOI)or,equivalently,iodinatedvolatileorganiccompounds(IVOCs).VOIsareemitted,
e.g.,bymacro-algae(Schalletal.,1994)aswellasfromphytoplankton(TokarczykandMoore,
1994;HillandManley,2009)andbacteria(Amachietal.,2001).Additionally,moleculariodine
(I2)isemittedfrommacro-algae(Küpperetal.,1998;Saiz-LopezandPlane,2004)afterreaction
ofhydrogenperoxidewithiodide.Thisresultsinformationofhypoiodousacid(HOI)whichisin
equilibriumwithmoleculariodineinseawater(Truesdaleetal.,1995).
Manybiogeniciodinecompoundsrapidlyphotolyseinthedaytimeatmosphereandrelease
atomiciodineradicals,e.g.(CH2I2+hν→CH2I+I)or(I2+hν→I+I).Assoonasatomic
iodineisavailable,reactionwithozoneformsiodinemonoxide:

I+O3→IO+O2(R20)
Inseveralcases,observationssupporttheconnectionbetweenground-basedmeasurementsofre-
activeiodinecompoundsandnearbyalgaeorphytoplanktoncolonies(Alickeetal.,1999).The
releaseofVOIsandmoleculariodinetotheatmospherebymacro-andmicro-algaethenleadsto
theavailabilityofreactiveiodinespecies.Incubationstudiesinthelaboratoryhaveconfirmedthe
emissionsofVOIsfromalgaetooccur,e.g.,asresultofoxidativestressuponmacro-algae(Pedersén
etal.,1996).Indirectcomparisons,polaralgaeseemtoexhibithigheremissionratesforiodinated
compoundsthansubtropicalalgae(GieseandWiencke,1999).Asimilarfindingwasmadefor
micro-algae,ascoldwaterdiatomswereidentifiedtoproduceiodinatedorganicsathigherrates
thantemperatespecies(TokarczykandMoore,1994;Mooreetal.,1996).
HighestatmosphericamountsofVOIswerereportedforCH3Iforindividualoccasions,onthe
EastcoastoftheUSAandattheAtlanticcoastinFrancewithlevelsashighas3800pptand
1830ppt,respectively(Lillianetal.,1975;Petersetal.,2005).FortheotherVOIsandmore
frequentlyalsoforCH3Ivolumemixingratiosontheordersof0.1-10pptwerefound,e.g.inthe
Antarctic(ReifenhäuserandHeuman,1992),attheIrishcoast(Carpenteretal.,1999)andat
HudsonBay(Carpenteretal.,2005).
Before2007,hardlyanyinformationwasavailableontheconcentrationsoreventhefluxesof
organoiodinesinSouthpolarwaters.RecentmeasurementsbyCarpenteretal.(2007)conducted

18

1.5Currentstateofatmosphericiodineresearch

intheAntarcticbetween70-72◦Sand9-11◦WinSouthernHemisphericsummer(December)then
showedthepresenceofCH2I2,CH2ICl,CH2IBrinthewatercolumnanddeterminedtherespective
fluxestotheatmosphere.ThisinformationisusedformodelstudiesinChapter5.Amountsare
rathersmallincomparisontomeasurementsinthemid-latitudes(Carpenteretal.,2001),butall
examplesrepresentpointmeasurementsunderspecificconditionsatrestrictedtimes.
Whilethedirectemissionofthesecompoundsfromalgaespeciesconstitutesanimportant
organicsourceofiodinecompounds,ithasalsobeensuggestedthatorganiccompoundsmaybe
releasedbyabioticpathways(Carpenteretal.,2005),followingthereactionofHOIwithhumic
material.Althoughorganicmaterialisinvolved,thisreleaseisnotanactiveorganicprocess.Con-
sideringpolyhalogenatediodocarbons,CH2I2andCH2ClIhavebeenobservedinfieldmeasurements
(Carpenteretal.,2005),aswellasCH2I2,CH2ClIandCHI3inlaboratorystudies(Martinoetal.,
2009).

1.5.2Troposphericiodinechemistryandozonedepletion
Fromthephotolysisoforganicorinorganicprecursors,atomiciodineisproduced.Atomiciodine
undergoesafastreactionwithozoneintheatmospheretoformiodinemonoxide(IO)andcatalytic
ozonedestructioncycleinareasofiodinereleasemaybeinitiated(ChameidesandDavis,1980;
.1994)al.,etSolomon

cyclesdestructionozoneCatalyticSeveralpathwaysmayregenerateatomiciodineincatalyticcycles,forexamplethroughthereaction
withotherhalogenoxides(X=BrorCl,R21,R22),oralternativelyHO2(R23)orNO2(R27)
photolysis.tsubsequenwith

I+O3→IO+O2(R20)
XO+IO→I+X+O2(R21)
X+O3→XO+O2(R22)
Net:2O3→3O2
I+O3→IO+O2(R20)
IO+HO2→HOI+O2(R23)
→OH+I+O2(R24)
HOI+hν→OH+I(R25)
OH+O3→HO2+O2(R26)
Net:2O3→3O2

19

bactificScien1kground

I+O3→IO+O2(R20)
IO+NO2+M→IONO2+M(R27)
IONO2+hν→I+NO3(R28)
NO3+hν→NO+O2(R29)
NO+O3→NO2+O2(R30)
Net:2O3→3O2
Thephotolysisreactions(R28)and(R29)inthelastcyclemayalsotakedifferentpathways.In
thosecases,nodestructionofozonetakesplaceasoddoxygenisrecovered.Thisalsohappensin
cycle.wingfollothe

lossozoneZero

Inthedaytimetroposphere,IOisalsophotolysedquickly,regeneratingIatomsbutalsoozone,
whichleadstosomesteadystateamountofIOduringdaytimeandtononetlossofO3:

I+O3→IO+O2(R20)
IO+hν→I+O(R31)
O+O2+M→O3+M(R32)
DuetothequickconversionbetweenIandIO,thetwocompoundsarefrequentlycombinedtothe
IOxfamily.AlthoughthephotolyticlifetimeofIOissmall(typicallyontheorderofminutesin
thedaytimeatmosphere),theeffectivelifetimeofreactiveIOxismuchlonger.

Influenceofiodineonothercompounds

Modelstudiessuggestthatevensmallamountsofiodinecanplayanimportantroleintherelease
andrecyclingprocessesofBratoms(Vogtetal.,1999).Thismayfurtherenhancethestrengthand
impactofbromineexplosionsseeninPolarRegions,andmakesthecatalyticozonedepletioneven
moreeffective.WhilethereactionofBrOwithitselfdoesnotproceedatrelevantrate,thecross
reactionwithIOleadstherenewedreleaseofatomicBr:

IO+BrO→OIO+Br(R33)
Thepresenceofiodineinthemarineboundarylayercanimpactontheratiosof[OH]/[HO2]and
[NO]/[NO2](ChameidesandDavis,1980).MainlythroughthereactionofIOwithHO2(R23),
the[OH]/[HO2]ratiomaybeincreased.SubsequentphotolysisofHOItoOHandIamplifiesthis
tendencyevenfurther.The[NO]/[NO2]ratioisusuallydecreasedinthepresenceofiodine,asthe
reactionofIOwithNOoccurs(ChameidesandDavis,1980).Reaction(R27)canshifttheeffectin
thereversedirection,buttheoverallbalancedependsontherelativereactionstrengths.

20

1.5Currentstateofatmosphericiodineresearch

1.5.3Higheriodineoxidesandparticleformation
Incoastalareas,timeperiodswithsuddenlylargeamountsoffineparticlesandaerosolsintheair
havebeenobserved(O’Dowdetal.,1999).Thesourceoftheseeventswasinitiallyuncertain.Nu-
cleationprocessesinvolvingH2SO4,H2OandNH3havebeendiscussed,butithasbeennoticedthat
anotherformerlyunknownspecieswasneededtoexplaintheparticlegrowthtothesizesdetected
inthemeasurements(O’Dowdetal.,1999).Marineaerosolhadbeenpreviouslyobservedtobe
enrichediniodinewithrespecttochlorineincomparisontotherelationinseawater(Duceetal.,
1963).Aerosolsthereforemightconstituteasinkofiodineundercertainconditions,butmayalso
actastemporaryreservoirsreleasingiodinespeciesbacktotheatmosphereatalaterpoint.Two
processesmayleadtohighiodineamountsinparticles.Iodinespeciesaretakenupbyalready
presentaerosolsandadditionally,theimportanceofiodinechemistryasasourceoffreshlyformed
particleshasbeenrecognised(O’Dowdetal.,2002b;Mäkeläetal.,2002).
Inadditiontofieldobservations,laboratorystudiesconfirmthecondensationofiodinevapoursto
solidspecies.Inthisprocesshigheroxidesofiodineplayacrucialrole.Higheriodineoxidesare,
e.g.,producedintheself-reactionofIO(CoxandCoker,1983;Blossetal.,2001;GómezMartín
etal.,2007).Tworelevantreactionpathwaysare:

IO+IO→OIO+I(R34)
IO+IO+M→I2O2+M(R35)
IodineoxidesmayalsobegeneratedbyreactionpathwaysbetweenIOandotherhalogenoxides
(ClOorBrO)orpossiblybyminorchannelsofthereactionofIOwithHO2:

IO+BrO→OIO+Br(R33)
HO2+IO→OIO+OH(R36)
TheselfreactionofOIO(O’Dowdetal.,1999;Jimenezetal.,2003)isrelevantandmightleadto
polimerisationandchainlikegrowthtoiodineoxideclustersinthefollowingform(O’Dowdand
2005):Hoffmann,

OIO+OIO→I2O4(or[IO]+[IO3]−)(R37)
I2O4+nOIO→[−I−O−IO2−]1+n/2(R38)
Severalaspectsarenotwellknownyet,includingadditionalpathwaysinthereactionsabove,the
ratecoefficientsandthestabilityoftheproducediodineoxides.Inanycase,higheroxides(IxOy,
suchasI2O4andI2O5)areformed,reactfurtherwithOIO,andformgrowingclusterswhichmay
subsequentlyprecipitate(O’Dowdetal.,2002b;McFiggansetal.,2004;O’DowdandHoffmann,
2005).Someofthehigheriodineoxideclusters,beingacidanhydrides,arehygroscopic.Theexact
mechanism,bywhichaerosolsareformed,isnotyetestablished.However,ithasbeenshownthat
theformationofhigheroxidesindeedresultsintheproductionoffineparticles.Thesemaythen
growandactascloudcondensationnuclei(O’DowdandHoffmann,2005),whichimpactsonthe
aerosolloadingandpotentiallyontheEarth’sradiationbudget.Iodineplaysanimportantrolein
both,thehomogeneousandheterogeneouschemistryofthetroposphereatleastlocally.

21

kgroundbactificScien1

Onemajorcontributortotheaerosolburdenoveroceanicregionsindependentlyofiodineis
sulphateaerosol(Brasseuretal.,1999).Therelativeimportanceofparticleformationfromiodine
oxidesdependsontheratioofiodinetosulphateparticles.Apartfromvolcaniceruptionsand
humanemissions,sulphurcompoundsareinjectedintotheatmospherebymarineplanktonmainly
intheformofdimethylsulphide(DMS,CH3SCH3)asreportedbyBatesetal.(1992).Viaoxidation
byOH,DMSformssulphurdioxide(SO2)andsubsequentlysulphuricacid(H2SO4).Homogeneous
nucleationofH2SO4andH2O(andNH3)mayleadtoadirectformationofparticles,whileH2SO4
alsocondensatesonpre-existingaerosols.DMSasprecursorofsulphateaerosolisconsideredan
importantsourceofcloudcondensationnuclei(CCN)inmaritimeregions(Charlsonetal.,1987),
whichaffecttheglobalradiationbudget.Itisanopenissuehowlargethepotentialimpactof
iodineparticlesonatmosphericCCNconcentrationsisincomparisontothesulphateaerosols.High
concentrationsofiodineparticleshavebeenobservedinseveraloccasions,e.g.byO’Dowdetal.
(2002a,b),forpointlocations,givingvaluableinformationontheprocessofparticleformationfrom
iodineoxides.Moreobservationsarenecessaryforthequantificationoftheiroverallimportance.
ObservationsofIOfromSciamachyarehelpfulinestimatingtheorderofmagnituderangeforthe
iodineparticleformation,whichmayserveasaglobalconstraintofitspotentialimportance(cp.
3.7).Sec.

1.5.4Iodineinthestratosphere
Asozonelevelsinthelowerstratosphereseemednottobeexplicablebythechemistryofchlorineand
brominealone,itwasthoughtthatiodinespeciesplayaroleforlowerstratosphericchemistry.The
influenceofiodineonstratosphericozonedestructionwasanalysedbySolomonetal.(1994)andwas
foundtobepotentiallylarge.Duetoveryefficientozonedestruction,iodinewouldneedtobepresent
onlyincomparablysmallamountsascomparedtochlorine,forexample.Solomonetal.estimated
thepotentialeffectofiodineonlowerstratosphericozonelosstobebyafactorof1000larger
ascomparedtochlorine.However,theoverallrelevancestilldependsontheactualiodinelevels.
Whileseveralstudieshavedemonstratedtheexistenceofiodineoxidesinthetroposphereabovethe
respectivedetectionlimits(cp.Sec.1.5.5),theabundancesofiodinespeciesinthestratosphereare
notyetwellknownandhaveoftenremainedbelowthedetectionlimits.
Themainreasonsforiodinecompoundstobelessabundantinthestratospherethanbromine
andchlorinespeciesaretheshorterphotochemicallifetimesandsmalleramountsofiodinated
precursorsubstances.Additionally,iodineisaccumulatedinaerosolswhichmayreducetheamount
ofgaseousiodinespeciesatleastinitially.TheWMOthereforestatesthat”itisunlikelythatiodine
isimportantforstratosphericozonelossinthepresent-dayatmosphere”(WMO,2006).
MeasurementsofstratosphericIOarerelativelysparse.Levelsofstratosphericiodinemonoxide
weremeasuredbyballoonbasedinstrumentation(Böschetal.,2003)insolaroccultationusingthe
DOASmethod.Theobservedmixingratiosremainedbelowthedetectionlimitof0.1pptbetween
12and20kmaltitudeintheprobedlocationsatseverallatitudesandtimesandalsoinsidethe
Arcticpolarvortexduringwinter.Attheseobservedupperlimits,theexpectedozonedestruction
isnotmuchlargerthanwithouttakingiodinechemistryintoaccount.
Recentfieldstudiesinthetropicssupportthefindingsoflowiodineburdensintheupper

22

1.5Currentstateofatmosphericiodineresearch

troposphereandlowerstratosphereregion,asupperlimitsforIOaswellasforOIOremainbelow
0.1ppt(Butzetal.,2009).Photochemicalmodellingleadstotheestimatethatiodinedoesnot
influencetheozonelevelsinthesealtitudesmuch.
Asimilarconclusion,thatthesignificanceofiodineforstratosphericozonelossismostprobably
negligiblewasalsodrawnbyWennbergetal.(1997)afterdetectingsmallamountsofIO(<0.5ppt)
byFourierTransformSpectroscopy(FTS)fromsunriseobservations.Someevidenceofstratospheric
IOwasreportedonthebasisofground-basedDOASobservationsinArcticregions(Wittrocketal.,
2000),identifyingpossiblemixingratiosbetween0.65and0.8pptinthestratosphere.
Duetolargeuncertaintiesinthemodelstudies,however,theestimatesoftheeffectofcertain
iodineamountsonstratosphericchemistry,especiallyonozonelevels,aretosomeextentuncertain
(Butzetal.,2009).Inaddition,thepotentialinfluenceofparticulateiodineisusuallynottakeninto
account,butiodinemaybereleasedfromtheaerosolphasebackintothegasphase.Consequently,
therelevanceofiodineforstratosphericchemistryremainsinpartanopenissueasitisnotclearly
supportedbyobservationsyet,butcannotbeexcludedeither.

1.5.5Observationsoftroposphericiodineoxides
Aftertheimportanceofiodinechemistryfortroposphericchemistryhadbeennoticedandestimated
in1980byChameidesandDavis,theinterestinmeasuringiodinecompoundsintheatmosphere
continuouslyincreased.Especiallythepotentialofiodineforozonedestructionisconsideredrele-
vantandhasbeenestimatedinseveralstudies.BeforeIOhadbeenreallyobservedintheatmo-
sphere(Alickeetal.,1999),iodinespecieshadbeenassumedtobepresentonlyinsmalleramounts
(ChameidesandDavis,1980;Solomonetal.,1994;Davisetal.,1996).Severalresearchcampaigns
havebeenconductedtocompletethepictureoniodineoxides.Theregionschosenforthesestudies
arefarspreadandcoverlocationsintheArctic,theAntarctic,severalmid-latitudecoastalsites,
andsomeothers.AnoverviewovertroposphericmeasurementsofiodineoxidesisgiveninTab.1.2.
In1999,Alickeetal.havedetectedIOmoleculesduringfieldmeasurements.Theymeasured
IOamountsinthemarineboundarylayer(MBL)atMaceHeadontheIrishcoastbylong-path
DOAS(LP-DOAS).ThisisanactiveDOAStechniqueusinganartificiallightsource,typicallya
xenonlamp(PlattandPerner,1980).Thelightbeamisguidedfromthelightsourcehorizontally
throughthesurfacelayersoveracertaindistanceoftypicallyseveralkilometersandisrecordedby
aspectrometerunit.Byusingaretro-reflectoratafarpoint,thelightemissionandlightdetection
unitscanbeplacedatthesamelocation.Themeasuredquantityisthevolumemixingratioofthe
respectivetracegasaveragedalongthehorizontallightpath.MaximumIOamountsseenbyAlicke
etal.wereintherangeof6ppt.Aconnectiontothetidalheightwasidentified.IOamountsincrease
forlowtideduringsolarillumination.Photolabileprecursorswerereleasedbynearbymacroalgae
whichareexposedtoairduringlowtide.Thisconnectionhasbeenobservedseveraltimessince.
Theinfluenceofiodineattheseconcentrationsonozonelevelshasbeenestimatedbymodelstudies
tobesubstantial(Alickeetal.,1999).ApartfromthemeasurementsbyAlickeetal.,morestudies
wereperformedattheIrishcoastfurtheranalysingandsupportingthecorrelationofIOamounts
withlowtide,highsolarilluminationandprecursoramounts(Carpenteretal.,1999;Allanetal.,
2000;Saiz-LopezandPlane,2004;Petersetal.,2005).AdditionalMBLsitesinthemidlatitudes

23

kgroundbactificScien1LocationMax.amountReference
IOMaceHead,Ireland2.4-7pptAlickeetal.(1999);Carpenteretal.(1999)
Allanetal.(2000);Saiz-Lopezetal.(2006)
Brittany,Atlanticcoast,France7.7pptPetersetal.(2005)
Roscoff,Northcoast,France30pptWhalleyetal.(2007)
Dagebüll,NorthSea,Germany1.9pptPetersetal.(2005)
Sylt,NorthSea,Germany2.2pptOetjen(2009)
Teneriffe,CanaryIslands3.5pptAllanetal.(2000)
DeadSea10pptZinglerandPlatt(2005)
CapeVerdeIslands1.2pptaverageamounts,personalcommunication+
GalapagosIslandsR.Volkamer,2009,amountsnotyetpublished
Maldives2.8pptOetjen(2009)
GulfofMaine,Atlantic4pptStutzetal.(2007)
HudsonBay<1pptHönningeretal.(2004)
Ny-Ålesund,Arctic0.4pptWittrocketal.(2000);Oetjen(2009)
CapeGrim,Tasmania2.2pptAllanetal.(2000)
NeumayerStation,Antarctica10pptFriessetal.(2001)
HalleyStation,Antarctica20pptSaiz-Lopezetal.(2007b)
OIO*MaceHead,Ireland3-9pptSaiz-Lopezetal.(2006);Petersetal.(2005)
GulfofMaine30pptStutzetal.(2007)
CapeGrim,Tasmania3pptAllanetal.(2001)
Tfromablea1.2:previousOvoerviewverviewofsomebyPimpetersortanettal.measuremen(2005).ts*Allanofioedinetal.oxides(2001)intheusetheatmosphere.absorptionUpdatedcross
sectionofOIOdeterminedbyCoxetal.(1999),whichisafactorof6largerthanthecrosssections
determinedbyBlossetal.(2001)andSpietzetal.(2005),usedinthestudiesfromPetersetal.
(2005),Saiz-Lopezetal.(2006)andStutzetal.(2007),withaccordinginfluenceonthederived
VMR.+unpublishedresults;http://www.chem.leeds.ac.uk/Atmospheric/Field/fage/iofw.html.
24

1.5Currentstateofatmosphericiodineresearch

wereexploredbyAllanetal.(2000),Petersetal.(2005)andOetjen(2009),reportingIOamounts
between2and8ppt.ThesestudiesusedtheLP-DOAStechniqueortheMulti-AXis-DOAS(MAX-
DOAS)method.Inthelatter,scatteredsun-lightDOASisperformedwithinstrumentsdesigned
toobserveradiationatamultitudeofdifferentelevationangles,therebypotentiallyyieldingsome
informationonthetracegasprofile(Wittrocketal.,2004;Heckeletal.,2005).
ThelargestamountsofiodineoxidesdetectedsofarhavealsobeenreportedforaMBLlocation
atRoscoff,situatedattheFrenchAtlanticcoast,byWhalleyetal.(2007)andWadaetal.(2007).
Whalleyetal.appliedtheLaserinducedfluorescence(LIF)methodforatmosphericIOmeasure-
ments.Inthisinsitumethod,IOmoleculesinarestrictedvolumeareexcitedwithaLaserbeam,
e.g.,withinthe445nmabsorptionbandoftheIOelectronictransition(cp.Sec.1.6.1).Theinten-
sityoftheinducedfluorescenceisobservedtodetermineIOmixingratios.Largestamountsaround
30pptwerereportedforcaseswhenashortintegrationtimeof10swasused.Atthesametime
andlocation,thefirstatmosphericmeasurementsofIObycavityring-downspectroscopy(CRDS)
havebeenperformedbyWadaetal.(2007).TheCRDSmethoddeterminesthedampingofalaser
pulsewithtimeinsideacavityenclosedbyhighlyreflectingmirrorsatbothends.Thestrongerthe
absorptionisinsidethecavity,theshorterthering-downtimebecomes.Thering-downtimesare
measuredonandoffresonancewithanIOabsorptionline,yieldingtherespectivemixingratioof
IOinsidetheprobevolume.Integratingover30sandprobingaconfinedairvolume,highestIO
amountsexceeding50pptwerefoundbyWadaetal..However,theCRDStechniqueforIOhasa
detectionlimitof10pptandhigherwithaccordinguncertaintiesontheresults.
ThestudiesusingLIFandCRDSnicelydemonstratethehighlyvariablecharacter,bothwith
timeandspace,ofIOamountsintheboundarylayer.Forlongerintegrationtimesorwithtechniques
averagingoverlargerspatialareasordistances,themaximumIOamountsaregenerallysmaller.
Theseobservationsinspatiallyconfinedlocationshaveimportantimplicationsfortheformation
offineiodineparticles.Thegenerationofhigheroxidesisnon-linearwithIOconcentrationas
discussedinSec.1.5.3.Consequently,therateofparticleformationincreasesstrongly,ifIOamounts
areconfinedtoasmallregionratherthanthesameamountofmoleculesbeingspreadoveralarger
area.ConcerningtheArcticPolarRegion,Wittrocketal.(2000)observedIObyground-basedDOAS
measurementsinNy-Ålesund,Spitsbergen.Althoughpartofthesignalwasassignedtosmall
amountsofIOinthestratosphere,thebehaviourofthedetectedabsorptionsignalwithrespectto
thesolarzenithangleindicatesthepresenceofsometroposphericIOinSpringtime.IntheSouth
PolarRegion,measurementswereconductedattheNeumayerStationusingscatteredsun-light
DOAS(Friessetal.,2001),andlateralsoatHalleyStation(Saiz-Lopezetal.,2007b)applying
LP-DOAS.ReportedIOmixingratiosareontheorderof10ppt.Theobservedseasonalcycleis
differentinthetwoAntarcticstudies.WhileFriessetal.(2001)findmaximumamountsinsummer,
theIOamountsseenbySaiz-Lopezetal.(2007b)arelargestinSpringtime(October),wheresingle
short-termamountsofIOreachupto20ppt.ComparingthetwoPolarRegions,theIOamounts
foundintheAntarcticaregenerallylargerthanintheArctic.Ground-basedobservationsinthe
CanadianArcticontheSoutheastcoastoftheHudsonBay(55◦N,75◦W),forexample,revealed
BrOatlevelsaround30ppt(Hönningeretal.,2004),whilenoIOabovethedetectionlimitof1ppt

25

kgroundbactificScien1wasfoundduringthesametimeperiod.
Inadditiontocoastalandpolarregions,furtherinterestinglocationsforboundarylayerhalogen
measurementsaresaltlakes.AttheDeadSeawithitshighsalinity,IOamountsaround10ppt
havebeenobservedbyZinglerandPlatt(2005).
TheOIOmoleculehassofarbeenobservedonlyinveryfewlocations,forexampleatMace
Headwithmixingratiosbetween3ppt(Petersetal.,2005)and9ppt(Saiz-Lopezetal.,2006)and
mostlyduringthenight.Recently,higheramountsof30pptandalsoduringthedaytimehavebeen
reportedbyStutzetal.(2007)fortheAtlanticcoastattheGulfofMaine,USA.
Thereareseveraladvantagestoeachoftheindividualmeasurementtechniques.MAX-DOAS
instrumentsdonotneedanartificiallightsourceandmaybeleftrunningself-containedalsoat
remotesitesforseveralyears.Additionally,theyyieldsomealtitudeprofileinformationforthe
analysedtracegas.Thispassivetechniquecanonlymeasureduringdaytime,whiletheother
techniquesarefunctionalalsoduringthenight.LP-DOASmeasurementsobservetheatmosphere
overalonghorizontaldistance.Signalsatseveralkilometresdistances,e.g.,abovetheocean,may
bepickedup.Forlocalisedsourceshowever,thesignalisaveragedoverthelightpathandtherefore
reduced,sothatnotthepeakconcentrationsaremeasured.LIFandCRDSbothconstituteinsitu
pointmeasurementsathightemporalresolutionandmaycapturepeakconcentrationsatconfined
locationsiftheyareemployedclosetosources,butdonotyieldanyinformationforthesurrounding
regions.LIFisaverysensitivetechniquewithtypicallyverylowdetectionlimits.
Inmostlocations,wherefieldcampaignswereconducted,theobservedIOandOIOamountslie
intherangebetween0and10ppt,whilehigheramountsareseldomandtransitory.Theground-
basedmeasurementsconstituteavaluablebasisofiodineoxideobservations.Theyare,however,
basicallylocalmeasurementsandcovershorttemporalperiods.Longtermglobalobservationsby
satelliteinstrumentsareadditionallydesiredtodeepentheinsightintoamounts,distributionsand
variationsofIOintheatmosphere.Fromthis,theunderstandingofsourceregionsandofthe
importanceofIOshallbeimproved.
1.6Atmosphericeffectsonradiation
Earth’satmosphereandwithitlifeonEarthisstronglyinfluencedbyanddependentontheincoming
andoutgoingradiation.ThesolarradiationreachingtheatmosphereoreventheEarth’ssurfaceis
aprerequisiteforlifeandresponsibleformanyprocesses,itdrivesphotosynthesis,initializesmany
chemicalreactionsbyphotolysisanddeterminestheairtemperature.TheSunemitsradiation
whichisclosetothatofablackbodyatatemperatureof5780K.Theirradiancethereforehasa
maximuminthevisiblewavelengthregionatabout500nm,whichmaybecalculatedfromWien’s
w:laλmax=2900μKm
TDuetoabsorptionandscatteringintheEarth’satmosphere,thespectralmaximumforthesolar
irradiancespectrumarrivingatthesurfaceisshiftedslightlytolongerwavelengths.Thephoton
numberfluxatthetopoftheatmosphere,however,hasitsmaximumataround580nmdueto
theinverselyproportionalenergy-wavelengthdependence.Asolarspectrummeasuredfromspace
26

1.6Atmosphericeffectsonradiation

bytheSciamachyinstrument(cp.Sec.1.9.1)isshowninFig.1.2.ThesharplinesintheSolar
spectrumaretheFraunhoferlinesandarecausedbyabsorbingspeciesintheSun’sphotosphereand
chromosphere.ThestrongestlinesaretypicallylabeledbyLatinletters(Stöcker,1995)asshownin
browninFig.1.2.TheFraunhoferG-bandlieswithintherelevantwavelengthrangeinsomecases
ofthepresentstudy.

KHGFEDC

Ca

FeNaH C&FeaH

Ca

Figure1.2:ASolarspectrummeasuredbySciamachycoveringallwavelengthchannelsofthe
Theinstrumenrightt.figureTheisspacloseectrumuphasofbtheeenspassemectrumbledbetwandeen300calibratedandb1000yJonmchenpointingSkupin,outIUPsomeofBremen.the
Fraunhoferabsorptionlineswithlabels(brown)andtheresponsibleatoms(blue).
Onitswaythroughthedifferentatmosphericlayers,theradiationisaffectedbyprocesseslike
scatteringatmoleculesandparticles,molecularabsorptionandreflection.Allthesefactorshaveto
betakenintoaccountinthecalculationofradiativetransfer.

rptionabsorMolecula1.6.1Theabsorptionofradiationintheatmosphereleadstoseveralfurthereffects.Forexample,alarge
portionoftheUVspectrumisabsorbedbyozonebeforereachingtheEarth’ssurfaceprotecting
lifefromtheharmfulradiation,andmoleculessuchasH2O,CO2,andCH4leadtothegreenhouse
effectbyabsorbing(andemitting)infraredandmicrowaveradiation.
Eachmoleculeandatomexhibitsitsuniquespectralabsorptionbands.Therefore,thedifferent
speciesmaybeidentifiedviaspectroscopicmeasurementsofthecharacteristicabsorptionstructures.
Theabsorptionofchemicalspeciesintheatmosphereaffectselectromagneticradiationpassing
throughregionswheretherespectivespeciesispresent.Byanalysingthisradiation,conclusionson
theamountofabsorbersubstancealongthelightpathcanbedrawn.Asacrucialprerequisitefor
theretrievalofatracegasamountbyopticalmeans,thewavelengthsofsomeabsorptionstructures
needtobecoveredbythemeasurementdevice.

27

tificScien1kgroundbac

Ingeneral,absorptionoflightisaccompaniedbytheexcitationoftheabsorbingmolecule.From
theinitialenergystateatransitiontoahigherenergyleveltakesplace.Threedifferenttypesof
transitionsaredistinguished,whichareelectronic,vibrationalandrotationaltransitions.Duringan
electronictransition,thequantumnumbersdescribingtheelectronicenergystateofthemolecule
change,i.e.,themolecularorbitalsoftheelectronsandtheoverallwavefunctionwillchange.
TheenergyofelectronictransitionscorrespondstoradiationintheUV,thevisibleorthenear
IRwavelengthregiondependingontheparticipatinglevelsandtheelectronicstructureofthe
respectivemolecule.Vibrationalorrotationaltransitionsdescribethechangeofvibrationalor
rotationalstates,vibrationaltransitionshaveenergiescorrespondingtoIRradiation,rotational
transitionscorrespondtowavelengthsinthemicrowaverange.
Ifthephotonenergyabsorbedbyamoleculeexceedsacertainlimit,thedissociationenergy,the
moleculemaybreakapartintotwofractions,i.e.,themoleculeisphotolysed.

Energystructureofadiatomicmolecule

TheenergystatesofamoleculearecalculatedfromtheSchrödingerequation(Demtröder,2000)with
severalmodificationsincomparisontoatomicenergystates.ThedifferentialSchrödingerequation
determinesthebehaviourandespeciallytheenergystatesofparticlewavefunctions,whichdescribe
theparticle’spropertiessuchasthespatialprobabilitydistribution.SolutionsoftheSchrödinger
equationareEigenfunctionsoftheHamilton-Operator,describingthetotalenergyofthesystem,
withEigenvaluesgivingthepossibleenergystates.
Foradiatomicmolecule,asetofquantumnumbersQspecifiestheenergyEigenvaluesEQ(r),which
arenotconstantasforatomsbutadditionallydependontheinternucleardistancerbetweenthetwo
involvedatoms.EQ(r)isamolecularpotentialcurveintheBorn-Oppenheimerapproximation,after
whichthemotionsofthecomparablyheavynucleicanbeseparatedfromthatoftheelectrons,and
thewavefunctionsoftheelectronsarecalculatedforagiveninternucleardistancer.Vibrationsand
rotationsofamoleculeinvolvethemovementofthenuclei.Thesemotionsneedtobeconsideredin
theSchrödingerequationandleadtonewquantumnumbersincomparisontoatomicstates.Which
quantumnumbersareusedforthedescriptionofthemolecularstate,dependsonthestructureof
themolecule,i.e.,onthestrengthoftheindividualmomenta.Importantquantitiesinthisrespect
aretheelectronicorbitalmomentum(asasumoverallelectrons)L,theelectronicspinS,the
nuclearrotationalmomentumRaswellasthetotalangularmomentumJ.Inadditiontothe
principalquantumnumbern,themagnitudesaswellastheprojectionsofthemomentaontothe
internuclearaxisyieldrelevantquantumnumbers.Theseare,e.g.,theprojectionΛofLandthe
totalangularmomentumJfrom|J|2=2J(J+1),with=2hπthereducedPlanckconstant.In
addition,υisusedtodefinethevibrationalstate.
Themolecularpotentialcandescribeboundstatesaswellasinstablemolecularstates,depending
onwhetherornotthecurveexhibitsanenergyminimum.Assoonasthenucleibegintomove
considerably,theirenergycontributestotheoverallmolecularenergy.

28

1.6Atmosphericeffectsonradiation

energyrationalVibThewavefunctionsofvibrationalstatesmainlydependontheshapeofthemolecularpotential.If
thiswasaparabola,thevibrationswouldbethatofaharmonicoscillator.However,theshapeis
notharmonicbutdistortedwithastrongerenergyincreaseforconvergingnucleithanwithgrowing
distance.Forr→0,theenergyneedstogrowtoinfinity,whileforr→∞,theenergyconverges
againstthevalueD,whichisthesumofthedissociationenergyandthezero-pointenergy.The
dependenceofthepotentialenergycurveV(r)ontheinternucleardistancerofadiatomicmolecule
canmathematicallybeapproximatedbytheMorsePotential,anempiricalformula,wherer0isthe
equilibriumbonddistance,andaaformfactor:

V(r)=D·(1−e−a(r−r0))2.(1.1)
Figure1.3showsaschematicoftheelectronictransitionsbetweendifferentvibrationallevelsina
diatomicmolecule.ThepotentialenergycurveswerecalculatedwiththeMorsePotentialformula.
ThepositionsofthevibrationallevelsandthedistancebetweenthetwoMorsepotentialcurvesare
notexactlytoscale,thevibrationallevelswouldlieclosertoeachother.Asthepotentialisnot
harmonic,thedistancebetweensubsequentvibrationalenergylevelsisnotconstantbutdecreases
withincreasingυ.ThesolutionsoftheSchrödingerequationfortheMorsepotentialyieldthe
energiesofthevibrationalstatesEvibwithangularfrequencyω:
2Evib=ωυ+1−2ω2υ+1(1.2)
2D42

energyRotationalIftheinternuclearaxisanddistanceriskeptfixed,rotationsaroundthisaxiscanbeconsidered
andtheenergiesErotarecalculatedinanalogytoclassicalrotationsfromthemomentofinertiaI
andthetotalangularmomentumJ:2
JErot=2I.
InsertingtheabsolutevaluesforJ2=J(J+1)2,andtheconstantvalueB=/(4πcMr02),with
thenuclearreducedmassM,thespeedoflightcandequilibriumdistancer0,theequationreads:

Erot=hcB·J(J+1)(1.3)
ThewavenumbersofrotationalemissionorabsorptionlineswouldincreaselinearlywithJfromthis
equation,andthedistancesbetweenadjacentlineswouldbeconstant.Inreality,theinternuclear
distanceslightlyincreasesduringrotation,therebyreducingtherotationalenergy.Therefore,the
energiesareslightlylowerthangivenbythesimpleapproximation.Energiesofrotationalexcita-
tionsarenotobserveddirectlyinthepresentstudy,asspectroscopyinthevisiblespectralregion
isperformed.Indirectly,rotationalenergylevelsplayanimportantrolefortheanalyseshere,as
rotationalRamanscatteringatN2andO2causesfilling-inofFraunhoferlinesinallDOASmea-

29

kgroundbactificScien1

3 '' =2 '' =1 '' =0 '' =

r0''r0'

' = 2' = 3
' = 0' = 1

2teelectronic sta

1teelectronic sta

DenergyDissociation

Figure1.3:Schematicenergybanddiagramofadiatomicmoleculeshowingtwoelectronicstates
withseveralvibrationallevelseach(distancesnotexactlytoscale).Theshapeoftheenergycurves
isdeterminedbytheMorse-Potentialformula,Eq.1.1.Thearrowsindicateabsorptiontransitions
ofthetypeusedforthedetectionofIOfromthelowestvibrationalstate(υ”=0)oftheelectronic
groundleveltothedifferentvibrationallevels(υ’=0,1,2,3...)ofthefirstelectronicallyexcitedstate.
Theequilibriumbondingdistancesaredenotedwithr0andr0forthegroundandexcitedstates,
.elyectivresp

surementsofscatteredsunlight.TheeffectofRamanscatteringonthemeasuredspectrawillbe
1.6.3.Sec.indiscussed

TheabsorptionspectrumofIO
Thespecificabsorptionbandsutilisedinthemeasurementandretrievalprocessbelongtothe
electronictransitionfromtheIOgroundstateX32/2todifferentvibrationallevelsofthefirstelec-
tronicallyexcitedstateA32/2.Therespectiveabsorptioncrosssectionforthistransitionisshown
inFig.1.4.Theshapeofthevibronic(vibration-electronic)absorptionlinesisinfluencedbythe
variousunderlyingrotationallines,whicharenotresolvedinthismeasurement.Theinvisiblesub-
structurecausesthelinestobroaden.Transitionprobabilitiesandalsotheoccupationnumbers
governedbytheBoltzmanndistributiondeterminethemagnitudeofthebands.
ThespectrumwasrecordedwithaFWHM(fullwidthathalfmaximum)valueof0.07nm(Spietz
etal.,2005;GómezMartínetal.,2005).TheIOabsorptioncrosssectionrevealsanextraordinarily
strongdifferentialstructureinthewavelengthrangearound400-460nm.Thepresenceofseveral
sufficientlynarrowpeaksinthecrosssectionofIOiscrucialfortheapplicabilityoftheutilised
spectroscopymethod.Fortheretrievalofatracegasamount,broad-bandabsorptionstructuresare
notconsidered,butonlycomparablynarrow(”differential”)featuresareanalysed.Themeasurement
techniquewillbedescribedindetailbelowinSec.1.8.Itisimportanttonotetheorderofmagnitude
ofthecrosssectionwhichis10−17cm2/molecule,andthereforeexceptionallylarge.Incomparison,
thedifferentialpartsofthecrosssectionsofNO2orO3obtainanorderofmagnitudeofmaximum

30

1.6Atmosphericeffectsonradiation

10−19cm2/molecule.IftheIOabsorptionwasonlythatstrong(weak),andgiventhetypically
observedamounts,itcouldnotbeobservedfromspacebythecurrentlyavailableinstrumentation.

6 0

0 5

300 4

20

10

00

GómezFigure1.4:MartínetMeasuredal.,2007)spwithectrumaofresolutiontheofabsorption0.07nm.crossSeveralsectionσtransitionofIObands(Spietzfromettheal.,ground2005;
statetothefirstelectronicallyexcitedstate(A32/2←X32/2)anddifferentvibrationalstates(υ←υ)
canbeidentified,causingastrongdifferentialstructure.

scatteringElastic1.6.2Radiationintheatmosphereisscatteredatmoleculesandparticles.Scatteringprocessesaredivided
intotwofundamentallydifferenttypes,i.e.,elasticscatteringeventsinwhichthephotonenergy
remainsunchanged,andinelasticprocesseswherethephotongainsorlosesenergyduringthe
teraction.inDependingonthesized(∼diameter)ofthescatteringobjectinrelationtothewavelengthλof
theincomingelectromagneticradiation,theelasticscatteringisreferredtoasRayleighscattering
(λd)orMiescattering(λ≥d).IntheRayleighapproximation,theshapeandmaterialofthe
objectdoesnotplayaroleandisneglectedinthecalculations.Thesizerelationinfluencesthe
phasefunctionofthescatteringprocess.Thephasefunctiondeterminesthedependencyofthe
scatteringprobabilityofthescatteringangle(θ),whichismeasuredwithrespecttothedirectionof
incomingradiation.ForRayleighscatteringincaseofanassumedpoint-likescatterer,therelation
simple:elyrelativisPRay(θ)=3·1+cos2(θ),
4whileforlargerandmorecomplicatedstructuresofthescatteringobjects,thephasefunctions
becomestrongerstructured.AlreadyforMiescattering,whichconsiderssphericalparticles,the
phasefunctionisnolongersymmetric,butforwardscattering(θ=0◦)ispreferred.
Thescatteringcrosssectionσscatofascatteringobject(atom,molecule,particle)isdefinedas

31

tificScien1kgroundbac

theratioofthepowerofthescatteredradiationperobjectandtheincominglightintensity.The
unitthereforeis[σscat]=m2.
TheRayleighscatteringcrosssectionforascatteringatomormoleculeforawavelengthλ=2πc/ω
isgivenby:44
ωeσRay=6πε02c4m2·(ω02−ω2)2+ω2γ2,
wheree,ε0,c,andmarephysicalconstantsdenotingtheelectricchargeunit,thedielectriccon-
stant,thespeedoflightandtheelectronmass,respectively.Theparametersγandω0describe
theattenuationcoefficientandtheresonancefrequency,whichdependontheindividualatomor
molecule.Fortypicalscatteringmoleculesintheatmosphere(N2,O2etc.),theresonancefrequen-
cieslieinthefarUVspectralrange,sothatforvisible(andnearUV)radiationtheapproximations
ωω0iswellfulfilled.Therefore,alltermsinthedenominatorexceptfortheconstanttermω04
canbeneglectedandthewavelengthdependencyofσRaybecomes:

4e4−4σRay=6πε2c4m2ω4·ω∝λ.
00Thiswavelengthdependencybecomesrelevantinseveralconsiderationsofatmosphericradiation
transport.Thescatteringcrosssectionoflargerandpossiblystructuredparticlesexhibitsaless
prominentwavelengthdependency.

scatteringRaman1.6.3Incontrasttotheelasticscatteringprocessesdescribedintheprevioussection,Ramanscatteringat
atomsandmoleculescausesawavelengthshiftofthescatteredphotonandtheprocessisinelastic.
Moreprecisely,aphotonisabsorbedbyanatomormolecule,whichpassesovertoavirtual(orreal)
levelforashorttime.Thestateofthemoleculethenrelaxesbackintoavibrationallyorrotationally
excitedleveloftheelectronicgroundlevel.Duringthisrelaxation,anewphotonisemitted.The
energydifferencebetweentheincidentandemittedphoton,consequently,isthedifferencebetween
thegroundlevelenergyandtherotationallyorvibrationallyexcitedlevel(seealsoFig.1.5).The
spectrumoftheoutgoingphotonsconsistsofseveralpeaksresultingfromthedifferentdiscrete
excitationlevels.ThesespikesarecalledtheStokeslines.Iftheatomormoleculeinitiallywasin
avibrationallyorrotationallyexcitedlevel,andtherelaxationleadsbacktothegroundlevel,the
emittedphotonhasahigherenergythantheincidentone.Theresultingspectrumconsistsofthe
so-calledAntistokeslines.ThelinestructureoftheRamanscatteringcrosssectionsforscattering
atN2andO2moleculesisdemonstratedinFig.1.6.Thepositionsandintensitiesaredetermined
bytheenergylevels(Eq.1.3)andtheindividualtransitionprobabilities.
Thisinelasticprocesshasimportanteffectsonthespectrumofscatteredsunlightascompared
todirectsunlight.Basically,scatteringofaphotonofacertainwavelengthoccurswithaprobability
proportionaltotheintensitypresentattherespectivewavelength.Duetothepresenceofstrong
andnarrowabsorptionfeaturesliketheFraunhoferlines,rotationalRamanscattering(RRS)at
airmoleculesstronglychangesthespectrumofscatteredsunlight.Inthesewavelengthregions,
thescatteringatthespectralsidesofFraunhoferlines,wheretheintensityisrelativelyhigh,is

32

1.6Atmosphericeffectsonradiation

Figure1.5:Thedifferencebetweenelasticscattering(A),wheretheoutgoingphotonhasthesame
energyastheincident,andtheinelasticRamanscattering,wheretheemittedphotonhaslongeror
shorterwavelengththantheabsorbedphoton(B).IntheStokescase,theatomormoleculeisleft
overinanexcitedvibrationalorrotationalstate(1b).Theinitiallevel(1a)isenergeticallylower.
IntheRamanprocesstheexcitedlevel(2)canbeavirtuallevel.

60

]240cm-30oioss SectCr01[n
200420

425Wavelength [nm]

ec-spRaman1.6:FiguretrumforN2(green)andO2
(blue)foranincidentwave-
lengthof425nmshowing
theStokes(λ>425nm)
O2andAntistokes(λ<425nm)
lines.Thespectrumwas
rogrammepausingcalculatedN2byMarcoVountasavail-
ttp://www.iup.hatableRing/AIR.huni-bremen.de/tml,∼vounseetas/also
430Vountasetal.(1998).

moreprobablethanforwavelengthswithintheFraunhoferlines.Therefore,acertainnumberof
photonsscattersfromthesidesoftheFraunhoferlinesintothewavelengthregionwithinthelines,
andfarlessphotonsfromwithinthelinesscattertolongerorshorterwavelengths.Asaresult,the
Fraunhoferlinesappearlessdeepandstronginthescatteredsunlightascomparedtodirectsun
light.CalculationsforthiseffectarepresentedinconnectionwiththeDOASretrievalmethodin
1.8.Sec.Theprocessoffilling-inoftheFraunhoferlineshasbeendiscoveredbyShefov(1959),Grainger
andRing(1962).ItwasthentermedtheRingeffectandwasexplainedonlylaterasresulting
mainlyfrominelasticRamanscattering(Brinkmann,1968).Thisfilling-inofabsorptionlinescan
beobservednotonlyfortheFraunhoferstructuresbutalsoforstrongabsorptionsintheEarth’s
atmosphere,suchasfromozone.Afterconsiderableabsorptionhastakenplace,e.g.withinthe
ozonelayer,andthepronouncedabsorptionbandsappear,inelasticRamanscatteringcausesa
noticeablefilling-inofthesestructures.Theexactshapeoftheresultingspectrumdependson
manyparameters,suchasthemainscatteringmoleculesandtheirenergylevelstructures.

33

Scien1kgroundbactific

TherotationalRamanscatteringonairmoleculesleadstowavelengthshiftsontheorderof
afewnanometers.Additionally,vibrationalRamanscattering(VRS)atliquidwatermolecules,
e.g.,inoceanstakesplace.Inlocationsoverwaterbodies,sunlightentersthewater,travelsa
certaindistanceandmaybescatteredbacktowardsthesatelliteinstrument.Onthewaythrough
theoceanwater,vibrationalRamanscatteringleadstoinfillingofabsorptionlinesduetothesame
principleasintheatmosphereandmayinfluenceatmosphericabsorptionmeasurements(Vountas
etal.,2003).ThedifferencebetweenRRSandVRSeffectsliesintheresultingspectralstructure,
asthetypicaldistancebetweenvibrationalenergylevelsleadstowavelengthshiftsontheorderof
nanometers.oftenseralsev

1.7Radiativetransferintheatmosphere

transferradiativeofDescription1.7.1Radiativetransferequations(RTE)calculatehowlighttravelsthroughtheatmosphere,considering
inprincipleallprocessesofabsorption,scattering,reflection,refractionandemission.Allradiation
inputandlossesatacertainaltitudeanddirectionneedtobecalculatedconsideringboth,direct
anddiffusepartsofthelight.Lossesarecausedbyabsorptionandscatteringofthedirectand
diffuseradiation,thesourceofradiationisthescatteringintotherespectivealtitudeandangleas
wellasthereflectionattheEarth’ssurface.EmissionprocessesarenegligibleintheUVandvisible
wavelengthregionsandthereforenotrelevantforthisstudy.Theseprocessesresultinthechange
oftheintensityofdiffuseradiationIwithaltitudez:

μdI(λ)=−I(λ,z)·ε(λ,z)+Q(I,λ,z)+R(I,λ,z)(1.4)
dzμdIdz(λ)=−I(λ,z)·σi(λ,z)·ρ(λ,z)+σRayρRay(z)+σMieρMie(z)
i+Q(I,λ,z)+R(I,λ,z)(1.5)

wherethesumrunsoverallchemicalspeciesi,andtheothervariableshavethefollowingmeanings:

I(λ,z)-radiationintensityatacertainwavelengthλandaltitudez
ε-extinctioncoefficientcombiningabsorptionandscatteringeffects
σi,σRay,σMie-crosssectionsofmolecularabsorption,RayleighandMiescattering
ρi,ρRay,ρMie-densitiesofabsorbingmolecules,RayleighandMiescatterers
μ-thecosineofthezenithangleθ
Q-integralsourcetermdescribingradiationgainfromelasticscattering
R-integralsourcefunctionforradiationfromRamanscattering
ThesourcetermQcontainsthreetermswhichdescribethesinglescatteringofthedirect
radiationfromabove,thesinglescatteringofthedirectradiationwhichhasbeenreflectedatthe
Earth’ssurface,andatermformultiplescattering,i.e.,scatteringofthediffuseradiationIk.For
thereflectionatthegroundinthesecondterm,aLambertiansurfaceisassumedwithaspectral

34

1.7Radiativetransferintheatmosphere

reflectanceα(λ)(albedo).Thesolarzenithangleθ0istakenintoaccountbyμ0=1/cos(θ0).For
simplicity,thewavelengthdependenceisdenotedbyasubscriptforthewavenumberk.

Qk(z,μ,φ)=bk(z)Fkexp(−τk(z))P(γ0,z,γ)
μπ40+bk(z)αkμ0Fkexp(−τk(0))dγP(γ,z,γ)exp(−τk(0)−τk(z))
4πμ0Ωμ
+bk(z)dγP(γ,z,γ)Ik(z,γ)
π4ΩThetwodirecttermscontainthesolarfluxFkatthetopoftheatmosphere(z0)andtheoptical
depthτk(z)oftheatmospherebetweenz0andz:zz0εk(z)dz.Theintegralsrunoverthesolid
angleΩandintegrateoverthecontributionsofdirectanddiffuseradiationscatteredaccordingto
thephasefunctionPfromzenithandazimuthangles(μ,φ)=γintoangles(μ,φ)=γ.Allsource
termsareproportionaltothealtitudedependentscatteringcoefficientbk(z).
TheRamanscatteringtermRcontainsthecontributions(energyredistributions)fromRaman
scatteredphotons.ThetermsmakingupRareequivalenttothetermsinQwithsomemodifications:
1.TheelasticscatteringcoefficientsbkneedtobereplacedbyrotationalRamanscattering
coefficientsbj,iRRSforascatteringprocessinvolvingawavelengthshiftλi←λj.
2.ForthecalculationofIkatλi,thedirectanddiffuseradiationatallneighboringwavelengths
λjneedtobeconsidered,andanintegration(practicallyasummation)overtherespectivespectral
erformed.pisregion3.ThephasefunctionPisreplacedbythephasefunctionofRamanscatteringPRRS.

DuetothedependenceofQ(andR)onthediffuseradiationI(z,γ)inthelastpartofmultiple
scattering,theRTEbecomesadifferential-integral-equation,forwhichnumericalsolutionmethods
arenecessary.Theradiativetransferprogrammingcodeappliedwithinthescopeofthisworkis
theSciatranmodel(Rozanovetal.,1997,2005b),whichhasbeendevelopedattheUniversityof
Bremen.

1.7.2TheSCIATRANradiativetransfercode
Sciatranisaradiativetransfercode,inwhichnumericalintegrationoftheRTE(Eq.1.4)is
implemented.Thecalculationcanbedoneindifferentdegreesofcomplexity,forexampleconsidering
theamountofinvolvedscatteringevents.Intheappliedmodelrunsinthisstudy,multiplescattering
istakenintoaccount.ConcerningthecurvatureoftheEarth’satmosphere,differentapproximations
arepossible.Inthefullsphericalmode,refractionintheatmosphereistakenintoaccount,while
thisisneglectedinthesimpleplane-parallelapproximation.Thepseudo-sphericalmodeconsiders
refractiononlyforthedirectradiation,butnotforthediffuseparts,andalsotakesintoaccount
thealtitudedependenceoftheradiationzenithangle.
DifferentversionsofSciatranareavailablewithindividualadvantages.Someversionscontain
improvementsaboveformerones.However,inthelatestversions,inelasticRamanscatteringis
notyetimplementedasanadditionaleffectonthetransportedradiation(neglectoftheterms

35

kgroundbactificScien1

containedinR).IntensityspectraincludingtheeffectofRamanscatteringneedtobecalculated
withSciatranVersion1.2.
ThesolutionoftheRTEusesthefollowingapproach.Inafirststep,avariableseparationis
performedfortheazimuthaldependenceoftheintensitiesandphasefunctions.Forthis,thescat-
teringphasefunctionsneedtobedevelopedintoasumofLegendrepolynomialsandtheintensities
arerewrittenasFourierseries.Furthermore,theapproachoffinitedifferencesisappliedwhich
transformstheintegralsintosums.GradientsofquantitiesdzdXwillbeturnedintofinitedifferences
ΔΔzXbetweentwoaltitudelevels.Inadditiontothealtitude,alsotheanglesarediscretised.Within
agivendiscretestepofthesevariablesthedependingfunctionsaretakentobeconstant.Thechosen
altitudegridwillinfluencetheaccuracybutalsothetimeconsumptionofthecomputations.
Beforeamodelruncanbestarted,theatmosphericstateandthegeometryofinterestneeds
tobedefined.Thisincludes,e.g.,theviewinggeometry(fromasatelliteorground-basedpointof
view),theviewinganglesandtheunderlyingsurfacealbedo.Aerosolproperties(models,amounts,
profiles)canbedefinedandaltitudeprofilesoftracegasesaswellastemperatureandpressure
profilesareneeded.Absorptioncrosssectionsofconsideredtracegasesandahighresolutionsolar
inputspectrumneedtobeprovided.Forthelatter,theFraunhoferatlasbyKuruczetal.(1984)is
used.

Withinthepresentstudy,theSciatrancodehasbeenusedforseveralcalculations:

•ThecomputationoftheRingeffect(cp.Sec.1.6.3)wasperformedusingSciatranVersion
1.2.AreferencespectrumdescribingthiseffectisneededfortheDOASretrievalroutineand
willbeexplainedindetailinSec.1.8.2.TodeterminethespectralstructureoftheRingeffect,
theintensityatthepositionofthesatelliteiscalculatedtwicebySciatran,onceincluding
without.onceandscatteringRaman

•Calculationsoftotalaswellasaltitudedependentlightpathenhancements(airmassfactors)
throughcertaintracegaslayersintheatmosphereareperformedusingSciatranVersion
2.0.Theconceptoftheairmassfactorisdescribedwithinthenextsection.Resultsofthe
calculationsarepresentedinsection2.3and2.4andusedforrelevantdiscussionpurposes.

1.8DifferentialOpticalAbsorptionSpectroscopy

Fortheretrievaloftracegasamounts,thetechniqueofDifferentialOpticalAbsorptionSpectroscopy
(DOAS)isausefulandwellestablishedremotesensingmethodwhichhasbeendevelopedand
improvedoverthelastdecades(Noxon,1975;PlattandPerner,1980;Solomonetal.,1987;Platt
andStutz,2008).Itwasfirstusedformeasurementsconductedwithground-basedinstruments,but
canalsobeappliedwhenobservingtheatmospherefromspace(Burrowsetal.,1999b).TheDOAS
methodmakesuseoftheindividualabsorptioncharacteristicsofmoleculesonthemathematical
basisofLambert-Beer’sabsorptionlaw:

36

I=I0·exp(−σρ·L).

(1.6)

1.8DifferentialOpticalAbsorptionSpectroscopy

I0andIarethelightintensitiesofadirectlightbeambeforeandafterpassingthrougharegion
oflengthLcontainingmoleculeswithabsorptioncrosssectionσandconcentrationρ.
Thefundamentaldifferencebetweentypicalradiativetransfercalculationsandthepointofview
intheDOASmethodisthewayhowthelightistracedthroughtheatmosphere.WhiletheRTE
considersanaltitudegridparalleltotheEarth’ssurfaceandintegratesoveralllossesandgainsat
eachgridpoint,theDOASequationwillinthebasicstepintegratetheradiationaffectingprocesses
alongtheindividuallightpath,whichmightbeinfirstplacelargelyunknown.Therefore,thetypical
verticalintegrationoverthealtitudezisreplacedbyanintegrationalongtheslantlightpathS
withincrementsds.Approximationsoftheaveragelightpathcanbedoneinasecondstep.This
way,thespectroscopicinfluencesonradiationandtheradiativetransferitselfareseparated.DOAS
typemeasurementscanbeconductedeitherusinganartificiallightsource(activeDOAS,e.g.with
xenonarclamp)orthesunasnaturallightsource(passiveDOAS).Inthelattercase,eitherthe
sunisobserveddirectlyortheatmosphereisviewedinadifferentdirection,monitoringscattered
sunlight.Inthepresentstudy,scatteredsunlightisusedaslightsource.Incontrasttotheactive
DOASandthedirectsunmethods,thelightpathhereiscomplexduetoscatteringevents.The
propertiesoftheradiationarrivingattheinstrumentisaweightedaverageoverallpossiblelight
pathsthroughtheatmosphere.ThestartingpointnowforthedevelopmentoftheDOASequation
reads:

dI(dsλ,s)=−I(λ,s)·ε(λ,s)

(1.7)

equationASDOThe1.8.1ForrealmeasurementsintheEarth’satmosphere,itismostlynotpossibletomeasureIandI0as
describedabove,wherethetwoquantitiesonlydifferintheabsorptionstructuresfromonespecial
tracegas.Usually,twomeasurementsarecomparedwherebothcarrymultiplespectralsignatures
fromatmosphericprocesses.Ideally,theI0spectrum,alsocalledthebackgroundspectrumor
generallythereferencespectrum,isatleastnotaffectedbythetracegasofinterest.
IntheDOASmethod,moleculesareidentifiedbythedifferentialpartσoftheabsorptioncross
sectioninsteadofbytheabsolutemagnitudewhichmakestheseparationofspectralabsorptionand
scatteringinfluencespossible.Allbroad-bandspectralstructuresareneglectedfortheidentification
processandonlythespectralfeaturesofhigherfrequencyareconsidered.Thespectralresolution
needstobehighenoughtoresolvethesestructures.Whilesomeotherabsorptionmethodsuse,
e.g.,onlythreedistinctwavelengthpositionsforanalysis(Breweretal.,1973),thepresentDOAS
methodworkswithawavelengthwindowatmoderatespectralresolutiontypicallyontheorderof
0.1-1nm.Allbroadbandinfluencesonthelightspectrumaremodelledbyapolynomialofsuitable
degree,aprocedurewhichcanberegardedashighpassfilteringofthespectrum.
Forthetaskofmeasuringatmosphericabsorbersunderatmosphericconditionsbyusingscattered
sunlight,theinitialsimpleformofLambertBeer’sabsorptionlawismodifiedinseveralaspects:

•Inmostcases,therewillbemorethanonesubstancepresentalongthelightpathwithab-
sorptionfeaturesinthesamewavelengthregion.Theextinctiontermhastobeextendedto

37

kgroundbactificScien1asumoverallrelevantabsorbersiwithindividualabsorptioncrosssectionsσianddensities
.ρi•Forscatteringatobjectsofdifferentsizes(i.e.,moleculesorparticles),thetwoscattering
modesRayleighandMiescatteringhavetobeconsidered(cp.Sec.1.6.2).Theirwavelength
dependencyisessentiallysmooth,sothatwhenconcentratingtheanalysisonasmallwave-
lengthwindowoftypicallyaround20-50nmwidth,thespectralinfluenceofscatteringcanbe
welldescribedbyapolynomial.
•Alsothebroadbandpartσboftheabsorptioncrosssectionsσ=σb+σwillbedescribedby
apolynomial.Alloccurringspectrallybroadbandfunctionsindependentoftheiroriginwill
besummarisedandapproximatedbyonepolynomialoforderp.
•Molecularabsorptioncrosssectionsnormallyarepressureandtemperaturedependent.With
changingaltitudeabovetheEarth’ssurface,therefore,thecrosssectionsofatmosphericspecies
mightchange.Thenmorethanonecrosssectionneedstobeconsideredpertracegas.In
somecasesthough,especiallyinthevisiblespectralregion,thiscanoftenbeneglected.
•Mostoftheparametersandvaluesintheequationarewavelengthdependent.Thisissome-
timesnotwrittenexplicitlybutneedstobekeptinmind.
ThesemodificationswillnowbeimplementedintheDOASequationandtheintegrationalongthe
lightpathSfromlocationL0toL1isperformed:
dIds(λ)=−I(λ)·σi(λ,s)·ρi(λ,s)+σRayρRay(s)+σMieρMie(s)
i⇒lnI|LL01=−ds(σi(λ,s)+σib(λ,s))·ρi(λ,s)+σRayρRay(s)+σMieρMie(s)
Sip
⇒lnII10=dsσi(λ,s)·ρi(λ,s)+akλk.(1.8)
S=0kiIncasetheabsorptioncrosssectionsarelargelyindependentofpositionstheycanbedetached
fromtheintegral.TheremainingtermissubstitutedbySC:=Sρi(λ,s)ds,whereSCiscalled
theslantcolumnandisgiveninmoleculespercm2.
plnI0=σi(λ,s)·SCi+akλk.
I1=0kiAstheaboveequationsonlyconsidertheidealcase,differencesbetweenmeasurementandtheory
cannotbecalculated.TheaboveconsideredmeasuredopticaldepthOD=lnII10needstobe
stillneedtobetakenintoaccount.Inanycase,themeasurementsareaffectedbynoise,which
replacedbythefitted,theoreticalopticaldepthODfitwhichdiffersfromthemeasurementbythe
left-oversignalsr(λ):
38

1.8DifferentialOpticalAbsorptionSpectroscopy

lnII10(λ)→ODfit(λ)
lnII10(λ)=ODfit(λ)+r(λ)
p⇒lnII10(λ)=σi(λ,s)·SCi+akλk+r(λ).(1.9)
=0kiTheslantcolumnamountsSCiandthep+1polynomialcoefficientsakarethemainretrieval
parametersbeingadjustedintheDOASfitroutinetoyieldthebestfitresultODfit,ascloseas
possibletothemeasuredOD.Thebestparametersaredeterminedinthefitroutinebytherequest:
Minimise(δ),withδ:=rj2
jThesumrunsoverallwavelengthpositionsjoftheinstrumentdetectorincludedinthespectral
fittingwindow.Theaboveequationrepresentsthequalitycriterionofaleastsquaresfit.The
differencespectrumr(λ)=(OD−ODfit)(λ)iscalledtheresidualspectrumandisofcentral
importanceintheimprovementprocessesofaretrieval.Thesuccessofaretrievalisinpartjudgedby
themagnitudeofδandtheappearanceoftheresidualspectrum.Smallvaluesofδandunstructured
residualspectrawithoutremnantsofabsorptionfeaturesorotherspectralstructuresarerequested
forasuccessfulretrieval.Inlaterdiscussionsofthefitqualitytheroot-mean-square(rms)ofthe
residualisusedasqualitycriterion,whichdescribesthedeviationofmeasurementandtheoryper
pixel,whereNisthetotalnumberofpixelsinthewavelengthwindow:
rms=1rj2.
Nj1.8.2TheRingeffectreferencespectrum
DuetostrongFraunhoferfeaturesintheUVandvisiblewavelengthregions,theRingeffecthasto
betakenintoaccountifscatteredsunlightisanalysedforatmospherictracegases(cp.Sec.1.6.3)
andturnsouttobecrucialfortheretrievalofiodinemonoxide.
AsrotationalRamanscatteringintheatmosphereleadstothefilling-inofabsorptionlines,the
effectontheresultingspectrumissimilartoanemissionprocess.Anadditionalreferencespectrum
σRRSdescribingthespectralfeaturesoftheRingeffectmaybeimplementedintheDOASretrieval
andusedinthesamewayastheabsorptioncrosssectionsσiinEq.1.9.TheRing-effectistherefore
alsoreferredtoasapseudo-absorber.
TheRingreferencespectrumcanbecalculatedusingaradiativetransfermodel.Inthepresent
study,theSciatrancode(Rozanovetal.,2005b)wasappliedandtherequiredintensityspectra
werecomputedfordifferentcases.DetailsonSciatranareprovidedinSec.1.7.Inthisradiative
transfermodel,thelightintensityatagivenwavelength(λ)travelingthroughtheatmosphereis
calculatedintwoways-withthepresenceoftheRamanscatteringeffect(I+)aswellaswithout
(I−)(Vountasetal.,1998).I+istheintensityoneactuallyrecordswhenobservingscatteredsun

39

1kgroundbactificScien

light.Allotheratmosphericeffectssuchastheabsorptionbyvarioustracegasesarealreadypresent
inthespectrumgivenby(I−).ThephysicalquantityunderinvestigationintheDOASmethodis
theopticaldepth.Ingeneral,withtheirradianceI0,thetotalopticaldepthτ+inpresenceofthe
Ringeffectisgivenby:

τ+=lnI+
I0−+II=lnI−+lnI0
=σRRS+τ−.

TheopticaldepthwithouttheinfluenceoftheRingeffectisτ−,andinthiswaytheRingeffectis
describedbytheadditivequantityσRRS:

+IσRRS=lnI−.(1.10)
InFig.1.7,theprocedureofgeneratingtheeffectiveRingspectrumisdemonstrated.Thetop
panelshowstwosimulatedintensityspectra,whichwherecalculatedwiththeSciatranradiative
transfercode,theblackcurverepresentingtheintensityspectrumneglectingtheinfillingbyRaman
scatteringandtheredcurvetakingthiseffectintoaccount.Fortheradiativetransfercalculations
analbedoof0.9,asurfaceelevationof0km(sealevel)and70◦SZAwerechosen.Fortestpurposes,
alsodifferentsettingswereexamined(cp.Sec.2.8).Thesmallerinsetmagnifiesthespectralregion
from415to440nmasthedifferencebetweenthetwospectraisrathersmallsothatitishardly
visibleinthelargegraph.Inthezoom-in,theinfillingoftheabsorptionlinearound430nm(the
FraunhoferG-band)canbejustrecognisedasthefeatureisslightlylesspronouncedintheredcurve
thanintheblackone(markedbygreycircle).TheFraunhoferG-bandconsistsofmanyabsorption
lines,mainlyoriginatingfromIron(Fe)andCalcium(Ca)ions.
ThelowerpanelshowstheresultingRingeffectcalculatedfromthetopcurvesfollowingEq.1.10.
Itbecomesclearfromthesefigures,thattheeffectissmallincomparisontothetotalintensities,
butintermsofopticaldepth,themagnitudeoftheRingeffectisconsiderable.Aswillbeshown
later,theRingeffectconstitutesoneofthemostinfluentialimpactsinthespectralregionoftheIO
retrieval.Theencircledfeaturearound431nmwillbeimportantforlaterdiscussion(cp.Sec.2.8).

1.8.3TheAirMassFactor

Iftheverticalcolumnofanatmosphericspeciesorevenitsconcentrationisneededratherthanthe
slantcolumn,someadditionalcalculationsandaprioriinformationarenecessary.Whiletheslant
columnistheintegratedtracegasamountalongtheactual,specificlightpath,theverticalcolumn
(VC)representstheverticallyintegratedtracegasamountfromEarth’ssurfacetothetopofthe
atmosphere.TherelationbetweenthesetwoquantitiesiscalledtheAirMassFactor(AMF)and
describesthelightpathenhancementascomparedtothedirectlightpathduetoslantirradiation
andobservationanglesandbecauseofscatteringprocesses.Itisnecessarilywavelengthdependent:

40

1.8DifferentialOpticalAbsorptionSpectroscopy

Figure1.7:TheintensityspectrainthetoppanelwerecalculatedusingtheSciatrancodeand
theyrepresenttheelectromagneticspectrabetween400and500nmastheywouldbemeasuredat
thepositionofSciamachy.WhiletheblackcurvedisregardsRamanscattering,thisisconsidered
intheredcurve.Ininsetzoomsinontheregionfrom415to440nm.TheRingeffectspectrum
determinedfromthesesimulationsisshowninthebottomgraph.Thegreycirclepointsoutthe
G-band.raunhoferF

AMF(λ)=SC(λ)
VCIftheAMFcanbecomputed,thentheSCcandirectlybetransferredintoaVC.Duetoscat-
teringincidencesthough,theaverageactuallightpathisusuallynotexplicitlyknown,sothat
thecalculationoftheAMFbecomeselaborateandneedstobeperformedusingradiativetransfer
dels.moRayleighandaerosolscatteringaswellassurfacereflectivityleadtothewavelengthdependency
oftheAMF.Thisdependencycanbeneglectedonlyasanapproximationinsmallwavelength
windowsandonlyforsufficientlysmallamountsofatracegas.Iftheabsorptionbyatracegas
amountbecomestoostrong,thentheAMFwillvarybetweenwavelengthswithstrongandweak
absorptionofthespecies.Inthatcase,thecomputationneedstobealteredtoincludeaniterative
algorithm.Forthecaseofiodinemonoxide,thisextensionisnotneeded.
TheAMFisinfluencedbythetracegasprofile,asthetypicallightpathenhancementchanges

41

bactificScien1kground

withaltitude.InthecomputationoftheAMF,therefore,anaprioriassumptionontheprofileshape
isneeded.Iftheprofileofthespeciesisnotknown,considerableuncertaintiesareintroducedinthe
processofconversionfromSCtoVC.Onlyverysparseinformationontheprofileisavailablefor
iodinemonoxide,sothatacalculationoftheVCbarestheriskofsystematicerrors.Nevertheless,
thecalculationofaprobableAMFmakessenseinordertoestimatethemagnitudeoftheVCand
alsoofthesurfaceconcentration.Oneshouldkeepinmind,however,thattheunknownprofile
shapehasasubstantialinfluenceandthatresultsintermsofVCorconcentrationsthatarederived
fromscatteredlightDOASmeasurementsaresubjecttocertainassumptions.

InadditiontothetotalAMF,thelightpathenhancementinacertainaltituderangeisof
interest.ThisisdescribedbytheblockAMF(BAMF)andcanberelatedtoanyaltitudeandlayer
thickness.TheBAMFisthediscretechangeinslantcolumndensityδSCioftracegasi,which
onewoulddetectwiththeappliedinstrument,iftheverticalcolumnchangesbyδVCi,jatacertain
altitudeintervalnumberjofthediscretealtitudegrid:
CδSiBAMFi,j=δVCi,j.
BAMFsarecalculatedinthisstudywithatypicallayerthicknessof200mintheboundarylayer
andincreasinglayerthicknessabove.Intheappliedradiativetransfercode,theBAMFiscalculated
viatheweightingfunction.Weightingfunctionsdescribethesensitivityofthemeasurementmethod
withrespecttoanyatmosphericparameter,e.g.theabsorptionoftracegases,independenceofthe
altitude.Theymaybedefinedindifferentways,andinthepresentcasetheweightingfunctions
describingtheintensitychange(δI)seenbytheinstrumentwithrespecttoachangeinmixingratio
(δVMR)inthealtitudeintervalhjareused,i.e.:
δIWI,i,jVMR=δVMRihj.
Asalinearchangeoftheintensitywithvaryingtracegasamountisassumed,thisapproachisonly
validforanopticallythinatmosphere.TheintensitychangeiscalculatedbyδI=−I0σiδSCi.
Usingtheaboveequations,thedensityofairρair,jataltitudej,andtheconversionoftheVMRto
verticalcolumnδVCi,j=δVMR·h·ρair,j,theBAMFscanbedirectlycalculatedfromtheintensity
weightingfunctions:WI,VMR
BAMFi,j=I0·σi,ji·ρair,j

AmeasurementisespeciallysensitivetoanatmosphericlayerweretheBAMFislarge,asthe
averagelightpaththroughthislayeriscomparablylongandtheabsorptionsignaturepickedup
bytheradiationduringitswaythroughthisspecificlayerisenhanced.Thisreducesthedetection
limit.AveragingtheblockAMFoverthealtitudeandweightingthesinglelayersbytherespective
relativetracegasamountsyieldsthetotalAMF.

42

1.8DifferentialOpticalAbsorptionSpectroscopy

1.8.4TheDOASfittingroutine
Afteratmosphericlightspectraarerecorded,thenextstepinobtainingthetracegasamountsfrom
thesemeasurementsistheDOASretrievalroutineitself.Forthispurpose,thefittingprogramme
Nlinisavailable,whichhasbeendevelopedattheInstituteforEnvironmentalPhysics,University
1997).ter,h(RicBremenTheprogrammeincorporatestheDOASretrievalmethodasdescribedaboveandcanbeused
fortheanalysisofsatelliteaswellasground-basedmeasurements.Thefittingroutineneedsspecific
parametersanddatafilesasinputinformationandyieldsoutputfileswiththeretrievalresults,
especiallythetracegasslantcolumnsofallchemicalspeciesincludedinthefit.Inadditionthe
spectralfitresultscanbesaved.Thestructureoftheprogramme,theincorporatedalgorithms,
routinesandsomeimportantcomputationalaspectsaswellasthenecessaryinputfileswillbe
wing.follotheinexplainedTheretrievalroutinestartswithreadinginallrelevantinputdata.Thisincludes,e.g.,the
measurementspectraIandreferencespectrumI0aswellasthelaboratorycrosssectionsσforall
includedtracegasabsorptions.Inaddition,aparameterfileprovidesdetailedinformationonthe
requiredfitparameters,suchasfittingwindow,includedtracegasspectra,polynomialdegree,file
pathsandotherretrievalsettings.
Aftertheinputdataisreadin,firstthespectralcalibrationoftheselectedbackgroundspectrum
I0isperformed.ThisstepusestheFraunhoferatlas(Kuruczetal.,1984),ahighresolutionspectrum
ofthesolarflux,andisgovernedbythestrongFraunhoferlines.ThespectralaxisofI0isadjusted
usingtwoparameters.Oneparametergeneratesaspectralshiftandthesecondfactormaystretch
orcompressthewavelengthaxis.TheFraunhoferreferencespectrumneedstobeconvolvedwith
theinstrument’sslitfunction.Theshiftandsqueezeparameters,apolynomialforthebroad-band
effectsandaroughcorrectionforthefilling-inoftheFraunhoferlines(theRingeffect)inthe
backgroundspectrumareretrievedbyusingthenon-linearLevenberg-Marquardmethod.
TheindividualmeasurementspectraareadjustedwithinthemainDOASretrievalasdescribed
below.Fortheiradjustmentalsotwoshiftandsqueezeparametersareapplied.Bothparameters
arenotallowedtoexceedaselectablelimit.Thewavelengthgridofthelaboratorycrosssections,
however,iskeptfixed.Shiftingandsqueezingisalsopossibleforthesespectraandwasappliedin
sometestcases,butwasnotusedforthefinalresultsinthepresentstudy.
Thetreatmentoftheabsorptioncrosssectionsisslightlydifferentforsatelliteandground-
baseddatahere.Forthesatelliteretrievals,laboratoryspectrameasuredwiththeinstrument
itselforspectrawhichareconvolvedwiththeappropriateslitfunctionpriortoread-inareused.
Fortheground-basedmeasurements,alllaboratorycrosssectionshavebeenmeasuredbydifferent
instrumentsandneedtobeconvolvedwiththeinstrumentslitfunctions.Forthiscase,adaily
measuredslitfunctionisread-in,andthecrosssectionsareconvolvedaspartoftheretrieval
routine.InthemainDOASretrievalprocesssolvingEquation1.9,firstthelogarithmofthemeasurement
dataisformed,calculatingtheopticaldepth.Thequantityδisthenminimisedinatwo-partfit.
Iteratively,anon-linearandalinearfitarerepeateduntilthealignmentofexperimentandtheory
judgedbythemagnitudeofδisoptimised.Anon-linearfitisperformedforthespectralalignmentof

43

kgroundbactificScien1

theindividualmeasurementspectrumwithrespecttothebackgroundspectrum.Foragivenchoice
ofshiftandsqueezeparametersforthecurrentmeasurementspectrum,alinearfitfortheslant
columnsandthepolynomialcoefficientsisperformed,followingEq.1.9.Theseretrievalparameters
aredeterminedsimultaneously.Bothsteps,thenon-linearandthelinearpart,arerepeateduntil
thealignmentdoesnotfurtherimprove.
Inadditiontothetracegasabsorptionspectrasomeotherspectraleffectsaretakenintoaccount:
1.TheRingeffectdiscussedaboveinSec.1.6.3andSec.1.8.2isconsideredbyusingapseudo-
absorptioncrosssectionσRRS(Eq.1.10)whichistreatedinexactlythesamewayastheabsorption
crosssections.ThefittingparameterintheDOASretrievalisthencalledafitfactorinsteadofa
column.tslan2.Straylightmayentertheinstrumentandaffecttheretrieval,astheDOASequationin
itsidealformdoesnotconsideranystraylight.AconstantamountCofstraylightchangesthe
opticaldepthtoreadln(II+0C).Thiscanberewrittenasln(II0)+ln(1+IC)andapproximatedby
ln(II0)+IC.Thesecondtermmaybetreatedasanadditionalcrosssection.TheconstantCis
chosenasthemaximumintensityinthefittingwindowtimes0.a03·Imaxcertainfactor,inthepresentcase
0.03.Theresultingreferencespectrumreads:σstray,offset=I.
AsecondstraylighttermtakesintoaccountalinearwavelengthdependencyofC,i.e.alinear
changeofthestraylightamountoverthedetector.Theopticaldepthbecomesln(I+CI+0D(λ)),with
D(λ)=0.03·Imax·λ2λ2−λforafittingwindow[λ1,λ2].Againafactorof0.03isapplied,whichhasin
bothcasesnolargeinfluence.Thisyieldsasecondreferencespectrum:σstray,slope=0.03·IImax·λ2λ2−λ.
Bothstraylight1effectspectraarethenscaledbyafitfactorretrievedintheDOASfit.Duetothe
dependenceof∼I,thespectralshapeissimilartotheRingeffectspectrum.
3.Furthermore,anundersamplingcorrection(Chance,1998)isincluded.Anarrowslitfunction
withasmallnumberofpixelsperFWHMmayintroduceerrorsintothefitinformofveryhighly
structuredspectralpatterns.Thismayoccurifashiftforthespectralalignmentisnecessaryinthe
fit.Afixedreferencespectrumisusedforthiscorrection.

instrumentsofDescription1.9

Severaldifferentinstrumentsandmeasurementdatahavebeenusedwithinthescopeofthisstudy.
Inthissection,overviewsanddetailsfortheemployedinstrumentsarepresented.Themainpart
oftheworkisconcernedwithobservationsfromtheSciamachysatellitesensor,whichwillbe
first.eddescribInfieldcampaigns,especiallyforvalidationpurposes(Sec.4)andforaspecificstudyfortheim-
provementofsatelliteretrievals(Sec.6),measurementsfromonecertaintypeofground-based(and
ship-based)instrumentationwereused.
Inaddition,furthersatelliteproductswereappliedforspecificdiscussionpurposesandaredescribed
attheendofthissection.

44

1.9.1ThesatelliteinstrumentSCIAMACHY

1.9Descriptionofinstruments

ThesatelliteinstrumentSciamachy(SCanningImagingAbsorptionspectroMeterforAtmospheric
CHartographY)ismountedontheEnvironmentalSatellite(ENVISAT)oftheEuropeanSpace
Agency(ESA).ENVISATwaslaunchedinMarch2002intoasun-synchronous,near-polarorbitwith
alocalequatorcrossingtimeof10amindescendingnode.TheobjectiveoftheSciamachymis-
sionistoprovideglobalmeasurementsofrelevantatmosphericcompoundsinordertoenhancethe
knowledgeandunderstandingofatmosphericprocessesandalsoofglobalclimatechange.Theanal-
ysisofdatameasuredbySciamachyyieldsinformationaboutEarth’satmosphere,includingthe
troposphere,stratosphereandalsothemesosphere.Theamountsofvariousatmosphericparame-
tersandcompounds,tracegasesandaerosolscanbedeterminedbyinversionoftheelectromagnetic
radiationspectrarecordedbySciamachy.Withthis,thechemicalcompositionandthepollution
oftheatmosphere,theozonechemistryandozonedepletioninthestratosphere,aswellasthesolar
variabilityandmanyotheraspectsarestudied.Theinstrumentalpropertiesandmissionobjectives
havebeendescribedindetailinBurrowsetal.(1995)andBovensmannetal.(1999).Anoverview
overtheinstrumentandsomemissionsuccessesaresummarisedinGottwaldetal.(2006).
Sciamachyisacombinedprismandgratingspectrometerandrecordselectromagneticradia-
tionfromscatteredsunlightinawidespectralrange(UV-vis-NIR-SWIR)andindifferentviewing
geometries.Innadirviewinggeometry,theinstrumentviewsdownwards-sunlightbackscattered
andreflectedfromtheEarth’satmosphereandsurfaceisrecorded.Measurementsfromthisview-
inggeometrywillbeanalysedinthefollowingchaptersforthespectralabsorptionsignaturesof
iodinemonoxide.Additionally,measurementsinlimbviewinggeometryareperformed,wherethe
instrument’sviewinganglepointsinflightdirectiontangentialtotheEarth’ssurfaceindifferent
altitudestepsinordertoresolveprofilesoftracespecies.Inthesolarandlunaroccultationmodes,
thedirectsunlightorsunlightreflectedatthemoon’ssurfacearerecorded.Combiningmorethan
oneofthesegeometriesallowsadditionalanalyses.Limbandnadirmeasurementscanbeutilised
togethertodeterminethetroposphericcolumnsofrelevanttracegases.Duetothealternatinglimb
andnadirviewingsettingsandaswathwidthof960km,globalcoverageattheequatorisachieved
withinsixdays.Foranimpressionofatypicalsequenceofnadirdata,Figure1.8displaysthe
locationsofnadirviewingmeasurementsforonesampleday,Sep1st2005.Alllocationsmarkedin
greenarewithinthefieldofviewofthenadirdataonthisday.Oneofthegreenmarkedareasis
alsoreferredtoasa”state”ofmeasurements.
TheopticalpropertiesaredeterminedbythedesignandcharacteristicsoftheOpticalUnitof
Sciamachy.AnoverviewoftheopticalsetupisprovidedinFig.1.9.InEarthobservingnadir
configurationtheradiationisguidedintotheinstrumentbytheElevationScanMirror(ESM).For
thesemeasurements,theoptionalApertureStopiskeptlargeandtheNeutralDensityFilteris
movedoutoftheopticalpathforoptimisedintensitythroughput.FollowingtheESM,thelightis
focusedbythetelescopemirrorontoentranceslitofthespectrometer.Determinedbytheoptics
andtheslitdimensions,theInstantaneousFieldofView(IFoV)hasasizeof1.8◦×0.045◦,i.e.,
approximatelyaninstantaneousgroundscenesize25km×0.6km(along-track×across-track).This
pixelsizeisenlargedduringmeasurementoperationbyscanningmovementoftheESMtoground
pixelsoftypically30km×60km.Thescancoversatotalswathwidthof960kmwhichmeans

45

kgroundbactificScien1

Figure1.8:Showningreenarethe
lomentscationsoncoonevdaeredy,bthey1nadirstofmeasure-Septem-
bcaler,daily2005,coveragedemonstratinganddatathetamounypi-t
ZenithSciamachyAngleisprovides.restrictedTheto<84Solar◦
here.

viewingwithinanglesof±32◦acrosstrack.
Aftertheradiationhasenteredtheinstrument,thespectraldispersionwithinthespectrometeris
achievedintwoconsecutivesteps.Prismsareusedforpre-dispersionofthespectrum,whichisthen
splitintoeightspectralchannelsandguidedtoeightindividualdetectorunits.Reflectiongratings
thenprovidetheappropriatefinaldispersion.Measurementsintheeightspectralchannelsare
thereforeperformedsimultaneously.Sixcontiguouschannelsarepresentbetween214and1773nm
andtwoadditionalshortwaveIRchannels(1934-2044nmand2259-2386nm).Siliconphotodiode
arrays(1024pixels)areusedinchannel1to5fordetectionofthephotonsignal,whiletheshort-
waveIRchannelsrequiretheuseofIndiumGalliumArsenide(InGaAs)asdetectormaterial.
Theindividualchannelsareagainsubdividedintoclusterswhichmayhavevaryingexposuretimes.
Thedivisionoftheeightchannels,theirspectralrangeandthetypicaltracegaseswhichcanbe
retrievedfromtherespectivemeasurementsaregiveninTab.1.3.Dependingontherespective
wavelengthregion,thespectralresolutionliesbetween0.2nmand1.5nm.Fortheretrievalof
iodinemonoxideinthispresentstudy,measurementsfromchannel3,specificallyfromclusters14
and15(404-527nm),innadirviewinggeometryhavebeeninvestigated.Inthischannelthe
spectralresolutionis0.44nm.

Inadditiontothemainchannels,SciamachycomprisesthePolarisationMeasurementDevices
(PMDs).TheoriginalpurposeofthePMDsistoprovidecorrectionsforpolarisationeffectsinthe
Sciamachysciencechannels2-6and8.WhilethemainSciamachymeasurementsrecordradiation
frombothpolarisationdirections,thePMDsaremainlysensitivetolightwhichispolarisedparallel
totheentranceslitofSciamachy.ThePMDsrecordtheintensityofradiationinsevendifferent
bands,involvingsixdistinctwavelengthintervalsgiveninTab.1.4.RecordingsfromthePMDs
becomeimportantforSec.2.7,astheycanprovideinformationoncloudcoverandtheunderlying
surfacetypeindifferentways.

46

toChannel 7

toChannel 8

fromSubsolar
Elevation Scanner

toChannels
fromNadir3-6
el 1Lev

el 2Lev

hCl 1nneal 2nneahC

AzimuthScanner
FromLevel 1
l 3nneahCl 4nneahC

1.9Descriptionofinstruments

l 5nneahC

Flight Direction-Y

-Z (toNadir)

pe

l 6nneahC

-X

Figure1.9:SchematicoftheopticalconfigurationofSciamachy.Thespectraldispersionis
achievedbyprisms(pre-dispersion)showninyellowinLevel1(top)andindividualreflectiongrat-
ings(blue)intheeightspectralchannels(channel3-6inLevel2,theothersinLevel1).Inthis
work,datafromchannel3innadirdirectionisused,whichisguidedintotheinstrumentbythe
ESMmirror.Thechannel3detectorunitispartofLevel2(bottomgraph).Figuresadaptedfrom
Gottwaldetal.(2006).

47

bactificScien1kground

ChannelNo.WavelengthIntervalTraceGases
ChannelChannel21300214--412334ONO3,BrO,SO2,OClO,HCHO
ChannelChannel43595383--812628HNOO2,O,O4,,NCOHOCHO,IO
322ChannelChannel65971773--17731063CO2,CH4
ChannelChannel7819342259--23862044CO

Tandablesome1.3:traceOverviewgasestoyverpicallytheretrievSciamaedchyinthismainspcectralhannels.windoForweacarehcgiven.hannel,Apartthewavfromelengththeseregiontrace
gasamounts,severalaerosolandcloudparametersmayberetrievedfromtheSciamachymeasure-
ts.men

PMDbandWavelengthintervall
21455-515310-365nmnm
43800-900610-690nmnm
652280-24001500-1635nmnm
7800-900nm(45◦polarised)

Table1.4:ListofchannelsandcoveredwavelengthbandsofthePMDsbelongingtoSciamachy.

systemsASMAX-DOGround-based1.9.2Multi-AXis-DOAS(MAX-DOAS)instrumentsareabletoviewlightindifferentelevationangles.
TheDOASinstrumentsoperatedbytheIUPBremenobservescatteredsun-lightandaretherefore
consideredpassiveDOASsystems,incontrasttoactivesystemsusingartificiallightsources.Spec-
traintheUV/visiblewavelengthregionsarerecorded.Severalstudiesofatmosphericconstituents
havebeenconductedwiththesesystems(Wittrocketal.,2000,2004;Heckeletal.,2005).Thebasic
DOASsetupconsistsofthreemaincomponents,thelightgatheringunit,awavelengthdispersing
elementandadetector.TheBremeninstrumentsreceivelightfromacustom-builttelescopewhich
focusestheincominglightintoanopticalfibrebundle.Thefibrebundlereducesthepolarisation
sensitivityofthemeasurements.Thelightisthenguidedtotheentranceslitofagratingspectrome-
ter(Czerny-Turnertype).Differentsizesandtypesofgratingspectrometersareused,typicallywith
afocallengthbetween275and500mmandplanarreflectiongratingswith300or600lines/mm.
TwodimensionalCCDcamerasareusedforradiationdetection,onedimensiongivingthespectral
information.Theseconddimensionisusuallynotusedforadditionalinformation,butthefinal
spectrumisreceivedbyintegrationovertheseconddimensionsignificantlyimprovingthesignalto
noiseratioofthemeasurements.ThesizeoftheCCDchip,thefocallengthofthespectrometerand
thegratingconstantdeterminethebandwidthofthesystem.Typically,bandwidthsbetween80nm
and300nmareachieved.Acomputerisusedtocontroltheinstrument’smeasuringsequenceand
storestherecordedspectraandcalibrationmeasurements.Forspectralcalibrationandslitfunction

48

1.9Descriptionofinstruments

Figure1.10:SketchoftheMAX-
DOASinstrumentsetupwiththe
mainparts:telescope,glassfibre
camerabundle,andspectrometercomputer.unit,Addition-CCD
ally,calibrationlampsareusedfor
wanationvelengthofthecalibrationinstrument’sandslitdetermi-func-
tion.ThefibrebundleisofY-form,
i.e.theincidentlightisdivided
intotwoequalpartsandisfedinto
twospectrometerunitsatthesame
Ftime.olkardSkWittroetchck,kindlyIUPproBremen.videdby

measurements,amercury-cadmium(HgCd)calibrationlinelampisinstalledinthetelescopebox.
AsketchofthegeneralinstrumentsetupisgiveninFig.1.10.
Thetelescopesystemdeterminestheviewingdirectionoftheinstrumentintwodimensions,
theazimuthalangleandthepolarangle(elevationangleorlineofsight,LOS).Here,theLOSis
measuredbyconventionwithrespecttothehorizon,sothathorizontalviewingcorrespondsto0◦
andthezenithisobservedataLOSof90◦.Operatedfromtheground,theBremenDOASsystems
observeatseveralelevationangles,anglesotherthanthezenithbeingselectedbyarotatablemirror
withinthetelescopebox.Typicalsequencesincludethezenith-skydirection,afixedelevationangle
at30◦abovehorizonandascanfromLOS0◦(horizon)in1◦or2◦stepsto15◦or16◦abovethe
horizon.Theexactsettingsdependonthesiteandobjectivesoftheobservations.Theviewing
geometriesfromgroundandthesensitivitiestowardsdifferentatmosphericlayersaredepictedin
Fig.1.11,showingthetelescope(TS)-spectrometer(SP)unitandlightpathsfordifferentsolar
positions,i.e.athighsun/lowsolarzenithangle(θ1)andforlowersun(largerSZAθ2).Lightis
scatteredintheatmospherebyparticlesandmolecules(SC).TwoimportantinfluencesofSZAand
elevationanglebecomeclearinthispicture:
1.WithincreasingSZA(θ1→θ2),thepathlengththroughastratosphericabsorberlayer(StAb)
increasesandwithitthesensitivitytowardstracegasesinthestratosphere.
2.Withlowerelevationangle(βa→βb),thepathlengththroughatroposphericabsorberlayer
(TrAb)increases,whilethelightpathsthroughthestratosphere(followingray1aand1b)remain
thesame.Thesensitivitytowardstracegasesinthetroposphereandespeciallyintheboundary
layeristhusenhanced.Thisfactleadstotheimportanceofthelowerelevationanglesadditionally
usedintheMAX-DOAStechniqueascomparedtozenith-skyDOAS.

A1.9.3instrumentssatellitedditional

SeaWIFSNASA’sSea-viewingWideField-of-viewSensor(SeaWIFS)onboardtheSeaSTARsatellitewas
launchedin1997.SeaWIFSrecordselectromagneticradiationineightdiscretewavelengthbandsof
20or40nmspectralwidthbetween400and900nm.Amongstotherbio-opticalproperties,infor-

49

Scien1kgroundbactific

AbSt

bATr

aTSbSP

a1

SC

2

b1

Figure1.11:SimplifiedsketchoftheMAX-DOASlightpathgeometriesatdifferentelevationangles
βa,babsorbandersat(StAb)differenfortaSZAgivθen1,2.elevLowationersunangle,(SZAwhile=aθ2l)owerleadselevtoationhigherangle(sensitivitβb)yleadsfortostratosphericenhanced
sensitivitytowardstroposphericabsorbers(TrAb).

mationontheoceanicchlorophyll-aconcentrationisretrieved.Inthedatasetusedforcomparisons
inthepresentstudy(Chapter3),theinformationisprovidedonagridwith9kmresolution.If
thefieldofviewpartlycontainsicecoveronEarth’ssurface,themeasurementisdiscardedfrom
theproductduetosaturation.DataandmapsfromSeaWIFSarefreelyavailablefromtheOcean-
Colorwebpage(http://oceancolor.gsfc.nasa.gov)fromNASA,theNationalAeronauticsandSpace
AdministrationoftheUnitedStatesofAmerica.

AMSR-E

TheAdvancedMicrowaveScanningRadiometerforEOS(AMSR-E)instrumentisinstalledon
theAQUAsatellite,whichisoperatinginspacesinceMay,2002,andbelongstoNASA’sEarth
ObservingSystem(EOS).AMSR-Eisapassivemicrowaveradiometer,recordingradiationatseveral
distinctmicrowavebands.Amultitudeofparametersisobserved,suchasseasurfacetemperatures,
iceconcentrations,atmosphericwatervapourandseveralothers.Inthepresentstudy,information
abouttheicecoverageonEarth’ssurfaceisusedfordiscussionpurposesinChapter3.Theanalysis
oficeconcentrationisperformedwiththe89GHzchannel(Spreenetal.,2008),andthedataversion
utilisedinthisstudyisthe”AMSR-EASI6.25kmSeaIceConcentrationData,V5.5i”.Theseaice
mapswereobtainedfromGunnarSpreenandLarsKaleschke(2008),InstituteofOceanographyat
theUniversityofHamburg,Germany,digitalmedia(ftp-projects.zmaw.de/seaice)inJune2009.

50

GOME-2

1.9Descriptionofinstruments

TheGlobalOzoneMonitoringExperiment-2(GOME-2)isascanningopticalspectrometercovering
theUVandvisiblespectralregionsfrom240-790nm.IthasbeenlaunchedinOctober2006onthe
MetOp-Asatellite(thefirstofitskindinaseriesofthree).GOME-2isasuccessoroftheGOME
instrumentstillinorbitontheERS-2satellite.ThemainobjectivesofGOME-2aremeasurements
ofO3andothertracegaseswhichareimportantforO3chemistry,aswellasseveraltracegasesof
airpollution.GOME-2isviewinginnadirgeometryandhasthebigadvantageofanear-global
coverageeverydayduetothewideswathof1920km.Thegroundresolutionof40×80km2is
similartothatofSciamachy.ThepresentworkusesresultsfromGOME-2inChapter6forthe
liquidwaterpathinoceanregions,aswellasatestmeasurementusinganewlygeneratedcorrection
spectrum.Infuturestudies,theretrievalofanIOproductfromGOME-2mightbeachieved.

51

2Developingtheretrievalofiodinemonoxide
satellitefrom

Untilrecently,measurementsofiodinespecieshaveexclusivelybeenconductedusingground-based
andinsomecasesballoon-borneinstruments.Theserepresentpointmeasurementsoftracegas
amountsatspecificlocationsandgivesomeinsightintothetemporalevolution.Informationon
aspatiallylargerscalemaybeobtainedfromaeroplaneorsatellitemeasurements.Theobjective
hereistousesatellitedataanalysistoobserveiodinemonoxidecolumnsfromspaceandtherewith
enhancetheknowledgeandunderstandingofiodineabundancesonamoreglobalscale.Thesatellite
sensorSciamachywasintroducedinSec.1.9.1andisofvaluablehelpinthisrespect.
Inthischapter,thetechnicaldetailsofthesatellitedataanalysisarepresented.Thedescription
startswiththenecessarydataprocessingsteps,beforethespecificDOASsettingsfortheIOretrieval,
thefitresultsandtheestimateddetectionlimitarediscussed.Dataqualityandconsistencyare
investigated.Thenpossibilitiesandeffectsofcloudscreeningarepresented,aswellasexample
influencescausedbycertainchangesintheretrievalsettings.
Selectedretrievalresultsandfirstexamplesofglobalmapsinthischaptershallspecificallyserve
fordiscussionpurposesoffitqualityandretrievalconsistency,whilethescientificdiscussionofthe
IOamounts,theirdistributionandapproachesforinterpretationwillfollowinthenextchapter.

2.1Satellitedataconfigurationandselection

BeforeiodinemonoxideslantcolumnscanberetrievedfromSciamachymeasurements,severaldata
configurationandselectionstepsarerequired.Thesinglestepsshallbespecifiedinthefollowing.

Level0toLevel1bProcessing

FromallavailablemeasurementsrecordedbySciamachy,acertainsubsetofdataisneededforthe
analysisofiodinemonoxide.Theinitialrawsatellitereadoutincountsperpixeliscommonlyreferred
toastheLevel0data.Firstofall,thisLevel0dataisprocessedbytheSciamachyDataProcessor,
whichusescontrolparameters,auxiliarydataandorbitparametersandyieldsthegeneralLevel1b
product.ThisLevel1bdatacontainsprocessed,geolocatedobservationaldatainscientificunits.
Additionally,instrumentmonitoringdata,calibrationmeasurementsandalsocalculatedcalibration
parametersareincludedinthisproduct.TheintermediatedataLevel1aisonlyaninternaldata
versionwithnopurposeforthedatauser.
Usually,satelliteinstrumentdataisdownlinkedfromthesatellitetoground-basedreceiver
stations,thenprocessedandtransferredtothescienceinstituteswithinseveralhours.Thisfirst

53

2Developingtheretrievalofiodinemonoxidefromsatellite

versionofthemeasurementdataiscalledthenearreal-time(NRT)data,andisthecommonly
usedversionforfirstscientificanalyses.Onlydatawhichisquicklyavailableaftertherespective
recordingiscontainedinthisNRTproduct.Consequently,someorbitsorsinglemeasurementscan
bemissinginthisversion.Asanothereffectfollowingfromthisimmediateprocessing,incaseofa
missingdarkmeasurement,thenextbestdarkmeasurementisprovidedinstead.Subsequentquality
analysisanddataactivitiesoftencompleteandimprovethedataset.Atcertaintimeslateron,so
calledconsolidatedandreprocesseddatasetsbecomeavailable,wheresomeformerlymissingorbits
aresuppliedandnon-optimalfirstchoicesforcalibrationsandcorrectionstepsarereplaced.The
advantageoftheconsolidatedandreprocesseddataistheusuallylargerdataamountandpossibly
ahigherdataquality.AclearadvantageoftheNRTdataisthealmostinstantaneousavailability
atanalreadygooddataquality.
FromthebeginningofthisstudyuptoMay22nd,2006,DataProcessorVersion5wasthe
implementedprocessingunitprovidingtheLevel1bproduct.Fordatarecordedlaterthanthis
date,anupdatedDataProcessorVersion6wasoperatedwithimprovedcalibrationstepsand
additionalchanges,e.g.inthesequenceofcalibrationandcorrectionsteps.Recently,thecomplete
SciamachymeasurementswerereprocessedusingthisrevisedProcessor(latestVersionnumber
6.03),fromwhichnowaconsistentLevel1bdatasetisadditionallyavailable.
Inthispresentstudy,twobasicdataversionsareapplied.Allscientificresultsandanalyses
aregainedonthebasisofnearreal-timedataforatimeperiodoffourandahalfyears,covering
thetimefromJanuary2004toJuly2008.Afterthereprocesseddataversionbecameavailable,
calibrationsteps(seeparagraphbelow)weretestedandalongerconsistenttimeserieswithslightly
differentcalibrationandcorrectionsettingswasgenerated.WhilefocusisontheNRTdatasetin
thisthesis,somenewanalysesandcomparisonshavealreadybeenconductedwiththereprocessed
data.Thetwodataversionswillbedescribedinmoredetailinthefollowingsectionsonfurther
dataprocessingandtheactualDOASretrievalprocedure.

Level1btoLevel1cProcessing
AsonlyacertainsubsetoftheavailableSciamachydataisneededforeachspecifictask,andas
individualrequirementsneedtobemet,theextractionandfurtherprocessingofdatafromLevel1b
toLevel1cisfacilitatedbyanextractiontoolprovidedbyESA.Level1cdatacontainsuserspecific
datasubsetswithindividualselectionandconfigurationsettings.Userscanselectthemselveswhich
properties(wavelengthregion,dates,viewinggeometry,calibrationsteps,etc.)areneededfora
ose.purpcertainForrecordingspriortotheprocessorchangeinMay2006,theextractionofSciamachyLevel
1bdatasubsetscanbeperformedwiththeSciaL1CCommandLineTool(Version2.5.0)fromESA.
ThistoolwaslinkedtotheDataProcessorVersion5,sothatitcouldnolongerbeusedafter
theprocessorexchange.Sincetheexchange,anupdatedSciaL1CTool(specificallyfordatafrom
ProcessorVersion6onwards,programmeRevision1.23)isavailable,whichmostlyworksinthe
samewayastheformerversion,butincorporatessomeupdatesonthecalibrationsteps.
Thedatautilisedinthisstudyfortheretrievalofiodinemonoxideisrecordedinnadirviewing
geometrybythethirdspectralchannelofSciamachy.Channel3issubdividedintoseveralclusters,

54

2.1Satellitedataconfigurationandselection

andthecurrentstandardIOfittingwindowfrom416to430nmoverlapswithclusters14and15,
bothbelongingtochannel3.
Consideringtheselectionofcalibrationanddatacorrectionsteps,formostpurposesofDOAS
retrievalsintheUVandvisiblespectralrange,ithasproventobethebestchoicetoselectonly
thedarkcurrentsubtraction(calibrationstep1)andtoomitallotherpossibilities.Usually,the
informationforwavelengthcalibration(calibrationstep5)isalsoextractedbutnotapplied(cp.
Sec.1.8).Thisinformationisonlywrittentothefileswithoutchangingthedataproduct.For
thereprocesseddataset,calibrationstep0forthecorrectionofthememoryeffectwasaddedto
theusualsettings.Thisstepappearstoyieldslightimprovementwithrespecttoearlierprocessor
versionsandalsoslightlybetterresultsthanwithoutthissetting.Therelevantdatacalibration
steps(programmeoptions”-cal0,1,5”)havethefollowingproperties:

0-cal•Thememoryeffectiscausedbyleftoverelectronsofaprevioussignalontheinstrument
detectoraffectingthesubsequentmeasurement,especiallyattransitionsbetweenbrightand
darksurfaces(suchasice/waterorcloudy/cloudfreescenes).Inthecorrectionprocedure,the
signalintensityoftheprevioussignaldeterminestheactionwhichisappliedtothesubsequent
readout.Measurementsfollowingaweaksignalarenotcorrected,whilefromrecordingsafter
intensesignalsaboveacertainthreshold,acorrectiontermissubtracted.Thiscorrectionis
calculatedasafunctionofthepreviousintensity.
1-cal•Thedarkcurrentisobtainedinseveralstatesonthedarksideofeveryorbitfromobser-
vationspointingintodeepspaceat250kmaltitudewherenoEarthshinelightisexpectedto
affectthemeasurement.TherearefiveDarkStatescoveringallrelevantexposuretimesused
formeasurements.Thecorrectionsignalinchannel3consistsofaconstantoffsetperreadout
andanadditionalpartfromtheleakagecurrentwhichisexposuretimedependent.
5-cal•Spectralcalibrationisperformedusingahollowcathodelamp(Pt/Cr-Ne).Thewavelength
axisiscalculatedfroma4thorderpolynomialfittedtothespectrallinesasafunctionof
detectorpixel.Thepolynomialcoefficientsarealsomodulatedbyaharmonicfunctionoforbit
phaseandfinallywrittenintotheorbitfilesofSciamachydata.However,thiscalibration
dataisusuallynotfurtherusedinthepresentstudy.Thewavelengthcalibrationisconducted
inadifferentway,asdescribedinSec.1.8.

Otherpossiblecalibration/correctionstepsimplementedintheextractiontoolarecorrections
forthepixel-to-pixelgain,theetaloneffect,theimpactofstraylight,polarisationinfluencesand
theradiometriccalibrationforcomputingtherealradiance.Insomequalitytests,theadditional
calibrationstepsdidnotimprovetheIOretrievalandwerethereforenotappliedtothedataused
.studythisinFigure2.1displaysatypicalspectrummeasuredbySciamachyinchannel3,hereshowingthe
spectralregionofcluster14andcluster15aftertheaboveprocessingstepshavebeenundertaken.

55

2Developingtheretrievalofiodinemonoxidefromsatellite

Figure2.1:Atypicalspec-
Scia-ybmeasuredtrummachyshownhereafterap-
plicationofthedarkcurrent
cali-ectralspandcorrectionfromdataincludingbrationcluster14and15inchannel
3.Thespectrumstwasmea-
suredonSep1,2005,with
agroundscenecoveringBre-
men◦ata◦centralpositionof
E.9.2N,52.8

Thedarkcurrentissubtractedandthespectrumisspectrallycalibrated.Somefeaturesattract
theattention.Atthechannelbordersthelightintensitydropsabruptlytozero.Inbetween,the
prominentdipsinthespectrumareFraunhoferlinesfromatomicabsorptioninthesolaratmosphere,
andthebroadintensityvariationwithaperiodontheorderof20nmiscausedbytheetaloneffect
ofthedetectorchip.SpectraofthistypeareutilisedasstartingpointfortheDOASretrievalof
.xidemonodineio

Dataadaptation,formattingandadditionalconditions
Usingdatafromtwoclustersatthesametime,herecluster14and15fromchannel3,attention
needstobepaidtotheintegrationtimesusedineachcluster.ThePixelExposureTime(PET)
isconstantfortheentirechannel,butforcluster14,thecoaddingfactorfcoadd,i.e.,thenumberof
subsequentrecordingsaddeduptooneaveragedmeasurement,andwiththistheintegrationtime
IT=fcoadd·PETislargerascomparedtocluster15.Thishasbeendecidedinordertoreducethe
dataamountduetothelimitedbandwidthforthedown-linkofdatafromthesatellitetoreceiver
stationsonground.Whenusingbothclusters,thecoaddinghastobematchedtotherespective
longerintegrationtimetoreceiveconsistentspectraacrosstheclusterborders.Thisadaptionand
aformatchangeareaccomplishedinanextstepbyaseparateprogrammewherealsothedata
formatisadaptedtotheinputformatrequiredfortheDOASfittingroutineprogrammeNlin(cp.
Sec.1.8.4).EachfileofsatellitedatacontainsoneorbitofSciamachymeasurements.Important
informationsuchasgeolocation,timeofmeasurement,viewinggeometry,appliedcalibrationsteps
andotherparametersaregivenforeachspectruminascii-headers,followedbytherespective
measurementspectruminbinarycode.
Duringorafterthefittingroutine,someadditionaldataselectionstepsareperformed.Thefollowing
settingshavebeenchosen:
•Thegroundscenesizeisincreasedtoaminimumof60kmalongand120kmacrosstrackby
averagingseveralindividualmeasurementsinordertoimprovethesignal-to-noiseratio.As
theintegrationtimeincluster14isusuallylongerthanforcluster15,thegroundscenesize
maybealreadyincreasedduetothematchingoftheintegrationtimes.

56

2.2TheDOASretrievalofiodinemonoxide

•Arestrictionforthesolarzenithangle(SZA)tolessthan84◦wasapplied.Thisexcludes
observationswithalowpositionofthesunandthereforeintrinsiclowsignal-to-noiseratio,
reducedsensitivitytothelowertroposphereandlargerretrievalerrors.

•TheslantcolumnsfromtheDOASfitresultaresubjectedtoadefinedqualitycriterion.Inthe
finalproductofatracegascolumn,onlythosedatawheretheresidualexhibitsasufficiently
smallroot-mean-square(rms)areincludedandutilisedforregionalorglobalmaps.

•Inthevisiblespectralregion,cloudsbetweenthesatelliteinstrumentandthepartofatmo-
sphereunderinvestigationcanaffectthetracegasretrievals.Inmanycasestheapplication
ofcloudscreeningalgorithmsisusefulornecessary.Informationaboutthecloudfractionina
satellitemeasurementcanbeobtained,e.g.,byaccessoryinformationfromotherinstruments
oranalyses.Elsewise,anintensitycriterioncanbeapplied,whichsimplyrejectsdatafrom
toobrightscenes.Inthepresentstudy,nocloudscreeningisappliedbydefaultforspecific
reasons,asexplainedlater.However,sometestsandcasestudiesforcloudscreeningwere
conductedtoinvestigatepossiblecloudeffectsontheIOretrievaleveninicecoveredregions
byusingarecentlydevelopeddataproduct.MoredetailsfollowinSection2.7.

Asummaryofallselectionandconfigurationparametersforbothdataversions-theNRT
productandareprocesseddataset-isgiveninTab.2.1.Amajorityoftheanalysesisconducted
ontheNRTdata,wheneverthereprocesseddatasetisused,thisisindicated.

2.2TheDOASretrievalofiodinemonoxide

AsdiscussedinChapter1,alltracegasesandatmosphericspectraleffectsaffectingthetransmission
oflightneedtobesimultaneouslyaccountedforintheDOASretrieval.Inthefollowingsections,
thespecificeffectswhichareofrelevanceinthespectralregionofIOabsorptionarepresentedand
theretrievalsettings,fitqualityandconsistencytestsarediscussedindetail.

2.2.1ThedevelopedIOstandardfit
Theretrievalwhichshowedthebestandmostconsistentresultsisreferredtoasthestandardfit.
TheresultsfromthisretrievalhavebeendescribedandusedinSchönhardtetal.(2008).Firstof
all,theparametersandsettingschosenforthiscasewillbementionedanddiscussed.Testswith
furtherconsiderationsoftracegasesanddifferentsettingshavealsobeenperformedandthemost
importantresultsfromthosetestrunswillbepresentedinSec.2.8.Oneimportantaspectinthe
followingisthejudgementofthequalityofafit.

LabelingdefinitionforIOproductversions
ApartfromthespecificstandardIOproduct,severalotherIOretrievalversionshavebeengenerated,
differingfromthestandardproductinseveralaspects,e.g.,intheSciamachydataversionorthe
choiceofthewavelengthwindowormore.Inmostofthefiguresandcalculations,resultsfromthe

57

2Developingtheretrievalofiodinemonoxidefromsatellite
ParameterSettingsNRTdataSettingsreprocesseddata
Processor(0-to-1b)Version5.0Version6.03
ExtractionTool(1b-to-1c)5.0V2.5.0and6.0Revision1.236.0Revision1.23
Calibration1and50,1,and5
(darkcurrent,spectralcalibration)(memoryeffect,darkcurrent,
calibration)ectralspandTimeinterval01/2004-07/200801/2003-02/2009
demonadirgeometryViewingSpectralregioncluster14,15inchannel3(404-527nm)
Spatialresolutionaveragedto60×120km2
◦4<8SZATable2.1:Overviewofrelevantconfigurationandselectionparametersappliedtothesatellitedata
scheduledfortheretrievalofIOslantcolumns.
definedstandardIOretrievalwillbeshownunlessstatedotherwise.Fortheappliedversions,ashort-
handnotationbecomesuseful.Inthisscheme,thestandardIOproductbasedonSciamachyNRT
datafromtheretrievaldefinedaboveislabeledversionV1.28.Asecondimportantdatasetis
basedonthereprocessedSciamachydatabutusesthesameDOASretrievalsettings,whereonly
minorchangesinthebackgroundspectrumwerenecessary.ThisversionisdefinedasversionV2.54.
ThefirstdigitmarkstherespectiveSciamachydatarelease(with1fortheNRTdata,2forthe
reprocesseddataset).Azeroasfirstdigitisusedfordatasetswhichhavenotpassedtheconsistency
checks.ThesecondnumberisarunningvaluefortherespectiveparametersetusedintheDOAS
retrieval.Inthisscheme,V0.27iisaproductshowingsystematicretrievalerrorsandwillbeused
fortechnicaldiscussionpurposesonly(cp.Sec.2.8).
settingsretrievalofChoicesTheabsorptioncrosssectionspectrumforiodinemonoxideexhibitsstrongdifferentialstructuresin
thewavelengthregionaround400to460nm.Inthisspectralregionadditionallyotherprocesses
alterthesolarradiationbeforebeingrecordedbythesatelliteinstrument.Theseeffectshavetobe
takenintoaccountinthefittingprocedure.
VersionSciamachydataIOretrievalRemarks
V2.54V1.28NRreproTdatacesseddatastandardstandard(cp.(cp.TTab.ab.2.3)2.3)newstandardproductproduct
telopmendevunderV0.27iNRTdatadiscussedalternativelatersettings(cp.Tab.2.7)withcorruptsystematicretrievalerrors,
osespurpdiscussionforusedTable2.2:Overviewoverthemainretrievalversionsrelevantforthepresentthesis.
58

2.2TheDOASretrievalofiodinemonoxide

Firstofall,theappliedIOabsorptioncrosssectionshallbeintroduced.Whileformostother
tracegasesdirectcrosssectionmeasurementswiththeSciamachyinstrumentareavailable,this
hasnotbeenperformedforminortracegasesasIO.However,laboratorymeasurementsoftheIO
absorptionhavebeenconductedathighspectralresolutionbyGómezMartínetal.(2007)and
Spietzetal.(2005)attheUniversityofBremen.Theassociatedtemperaturefortheabsorption
crosssectionis298K.ThistemperaturediffersfromthetemperaturesinsomelocationsonEarth
whereIOhasbeendetected.Theimpactisasourceofsystematicerror,butcanbeconsidered
minorincomparisontoothers(cp.Sec.2.5).ThelaboratoryabsorptionspectrumσIO(λ)needsto
beconvolvedwiththeSciamachyslitfunctionfSCIAbeforeenteringtheDOASfit.Thisprocedure
isdemonstratedinFig.2.2,resultinginthecrosssectionspectrumσIO,SCIA(λ)whichisusedin
thefollowingretrievalsforIO:
λ+ΔλσIO,SCIA(λ)=σIO∗fSCIA(λ)=σIO(λ)·fSCIA(λ−λ)dλ
λΔ−λTheRingeffectintroducedinSec.1.6.3isincludedinallretrievalruns.TheimpactoftheRing
effectonthemeasuredspectrumissubstantialfortheviewinggeometryandthewavelengthregion
here.AsafocusliesontheAntarcticregion,thestandardRingspectrumusedinthepresentstudy
iscalculatedforasolarzenithangleof70◦andagroundspectralreflectanceof90%.Twotrace
gasesthatneedtobeconsideredintheIOretrievalarenitrogendioxide(NO2)andozone(O3).
TheabsorptioncrosssectionsofbothspeciesweremeasuredbyBogumiletal.(2003)atdifferent
temperatures,fromwhichthespectraat223Kwerechosenhere.ThemajorpartoftheO3resides
inthestratospherewheretemperaturesareratherlow.Inremotesites,asimilarconsiderationholds
forNO2astroposphericamountsaresmall.ThechangingabsorptioncrosssectionofNO2for
highertemperatureswasalsoconsideredintestretrievalswithnolargeeffectontheIOcolumn.
ThespeciestakenintoaccountinthecurrentstandardfitarelistedinTab.2.3.Referencesto
therespectivelaboratorystudiesofthecrosssectionmeasurementsareincluded.Theparticular
absorptioncrosssectionspectraandtheopticaldepthoftheRingspectrumareshowninFig.2.3,
wheretheycanbedirectlycompared.
Apartfromthechoiceoftheabsorptionfeaturestobeconsidered,severalfitparametershavetobe
.adequatelyadjusted

Figure2.2:ThehighresolutionIOspectrum(left)isconvolvedwiththeSciamachyslitfunction
(greencurveinthecenterfigure).TheresultingIOspectrumwithslightlybroaderabsorptionpeaks
andsmallermaximaisthenusedintheDOASretrievalsofIOfromSciamachydata.

59

2

elopingDev

the

alretriev

of

dineio

xidemono

from

satellite

Figure2.3:SpectraoftheabsorptioncrosssectionsofO3,NO2,andIOandtheeffectiveoptical
depthoftheRingeffect.Whiletheabsorptioncrosssectionsareresultsoflaboratoryexperiments
(seereferencesinTab.2.3),theRingspectrumfollowsfromtheoreticalconsiderations.

60

2.2TheDOASretrievalofiodinemonoxide

•Broad-bandinfluencesareaccountedforbyapolynomialwhichissubtractedfromthemea-
suredopticaldepth(cp.Sec.1.8).Thepolynomialdegree,whichyieldsthebestretrieval
results,dependsonthesizeandpositionofthewavelengthwindowandalsoonthetrace
gasinquestion.Asecondorderpolynomialturnedouttobesuitableforthefinallychosen
standardfittingwindow.Intestfits,alsohigherorderpolynomialswereused,butthefitting
windowof416-430nmisrathershort,soalowdegreeisappropriatehere.
•Forthecorrectionofpossibleresidualstraylightintheinstrumentduringmeasurements,
aspectrallylinearstraylightintensityisassumed.Thestraylightreferencespectraare
generatedfromtheactualmeasurementbyusingfactorsof0.03fortheoffsetandslope
parameters(cp.Sec.1.8).Theresultingspectraarescaledduringthefittingprocedureinthe
samewayasabsorptionspectra.
•Anundersamplingcorrection(Chance,1998)isincluded.ThisisusefulincaseofScia-
machyretrievals,duetoarathernarrowslitfunctionandlowsamplingperFWHM.This
cancausefastvaryingfeaturesattheshouldersofspectrallines.Theresultinginfluenceson
themeasuredspectraneedtobecorrected.
•AnotherimportantchoiceinaDOAS-Fitisthebackgroundspectrumtowhicheachindividual
measurementiscompared.InthestandardIOfit,anEarthshinespectrumischosenas
background.ThischoiceispreferredoverasolarmeasurementinordertominimisetheRing
effectandresidualinstrumentaleffectscausedbydifferencesintheviewingofEarthandSun.
ThereferenceregionwasselectedinthePacificat40◦Sand160◦Wandwithin±10◦inboth
directionsasillustratedinFig.2.4.Thislocationisspecificallychosenasaregionwhere
theIOsignalisexpectedtobesmall,whichwasconfirmedbyfittingtheEarthshineagainst
thesolarreferencespectrum.ResultingIOcolumnsfromtherespectiveDOASfitarethen
differencesbetweenthecurrentmeasurementandtheaveragedreferenceregion.Effectively,
theIOamountinthereferenceregionissettozero.Theprocedureisimplementedasfitwith
respecttoaconstantEarthshinespectrumandsubsequentsubtractionofthedailyaverage
withintheassignedreferenceregion.ThisisthesameasusingthedailyaverageEarthshine
spectrumfromthereferenceregion.FurtherconsiderationsarepresentedinSec.2.9.

Figure2.4:Globalmapshowingtheref-
erenceregionovertheSouthernPacificat
40◦Sand160◦Wwithasidelengthof20◦
ineachdirection.Thisbackgroundregion
isusedforallVersion1data,e.g.the
V1.28.alretrievstandard

61

2Developingtheretrievalofiodinemonoxidefromsatellite

effectInfluencingabsorptionIOabsorptionNO2absorptionO3scatteringRamanRot.

ReferencetsCommenlaboratoryspectrum,293KGómezMartínetal.(2007)
laboratoryspectrum,223KBogumiletal.(2003)
laboratoryspectrum,223KBogumiletal.(2003)
Sciatrancalculation,(Sec.1.6.3)Vountasetal.(2003)

Settingsnm430-416ndorder2

SettingsparameteralRetrievWavelengthregion416-430nm
High-passfilterpolynomial2ndorder
correctiontlighyStra0.03parameteroffset0.03parametereslopBackgroundspectrum(I0)constantEarthshinespectrum
Table2.3:OverviewoftheretrievalsettingsvalidforthecurrentstandardfittingroutineofIO
columnsfromSciamachy,asappliedforproductversionV1.28.Toppanel:Absorptioneffectsby
tracegasesandRamanscattering.Bottompanel:Additionalretrievalparametersettings.

2.2.2FitqualityandconsistencyoftheIOstandardfit
Thequalityofaretrievalneedstobejudgedindifferentways.Apartfromcheckingtheresults
forconsistencyandforabsenceofartefacts,oneimportanttestforthefitqualityisthemagnitude
oftheresidual.Onlyiftheremainingresidualisreasonablysmall,thefittingprocedurehasbeen
successful.Additionally,itisnecessarythattheresidualspectrumisalsofreeofregularstructures,
butmostlyconsistsofanoisesignal.Therefore,ifdifferentretrievalsettingsarecompared,an
improvedfitshouldrevealareducedandlessstructuredfittingresidual.

JudgingthequalityofasatelliteDOASretrieval
WhentheoverallproductqualityofaDOASretrievalfromsatellitedataisassessed,thereare
severalconsiderationsthatshouldbetakenintoaccount.Firstofall,theretrievalqualityitself,
i.e.,therms-valueisimportant(cp.Sec.1.8).Whenthefitisimproved,themagnitudeoftherms
willdecrease,butthisdoesnotsufficeasaqualitycheck.Otherinspectionsarenecessary,which
aremostlyconsistencychecks:
-Physicallynotlogicalresults,e.g.,negativetracegascolumns,needtobeavoided.Thisapplies
tothetracegasinquestion,butalsotoallotherincludedabsorbersandeffectsinthefit.
-Theresultsfortheothertracegasesincludedinthefitroutine,whicharealreadywellknown,
needtobereasonable,i.e.reasonableamountsandtheknowntendencies,spatialvariations,special
regionsetc.shouldbereproduced.
-Theappliedcorrectioneffects,especiallytheRingeffectinthepresentstudy,needtoreceivethe
correctalgebraicsigninthedifferentregionsandreasonablemagnitudes.
-Anotherimportanttestinvolvesadditionalmeasurements.Goodagreementwithground-based

62

2.2TheDOASretrievalofiodinemonoxide

measurementsandresultsfromotherstudiesisdesired.Thisrequiresavailablestudiesforcompar-
isonandisdiscussedseparatelyinChapter4.

Theseconsiderationshavebeentakenintoaccountwhenjudgingthefitqualityofthedifferent
IOretrievals.Accordingtotherespectiveresults,thecurrentstandardIOretrievalperformed
satisfactorilyandexampleresultsoftheperformedtestsarepresentedinthefollowingparagraphs.

TwoexamplefitresultsfortheIOabsorption
FromtheaforementionedstandardIOfitV1.28,examplefitresultsaredisplayedinFig.2.5fortwo
individualSciamachymeasurements.Bothspectrawereselectedfromorbitnumber20051001_110
onthefirstofOctober,2005,ontheSouthernHemisphere.Exactdataisgiveninthefigurecaption.
ThetwoexamplesarechosenfortwodifferentamountsofIO,i.e.,2.1×1013molec/cm2(a)and
1.3×1013molec/cm2(b).Thefigureshowsthedifferentialabsorptionofthespecificmeasurement
atthetopintermsofopticaldepth.Thisspectrumstillincludesallabsorptionfeatures,butthe
fittedpolynomialhasalreadybeensubtracted.Thebottompanelshowsthefinalresidual,where
allassignedeffectshavebeeneliminatedfromthespectra.Onlytheunassignedstructuresremain
anditisvisible,thattheresidualdoesnotexhibitanylargeleftoverabsorptionfeaturesand
mainlyconsistsofmeasurementnoise.Thermsvaluesoftheresidualgiveninthefigurecaption
areveryclosetothetheoreticallycalculatedlimitwhichispossiblewithSciamachyrecordingsin
thiswavelengthregion.ThethreegraphsinthemiddleshowtheretrievalsofNO2,theeffective
Ringstructure(sumofRingandstraylighteffects)andIO.Thebluesolidlinerepresentsthe
scaledreferencespectrum,whilethereddottedlineisthereferencespectrumplustheresidual
structure.TheretrievalsshowthatIOisdetectedinthesecasesandthattheresidualdoesnot
containnoticeablesystematicstructures.Thisgivesconfidencetothechosenfitparametersettings.

ResultsforNO2intheIOstandardfit
TheIOretrievalsettingsarenotoptimisedfortheretrievalofNO2ofcourse,butnevertheless
thefitsoftheotherincludedtracegasesapartfromIOneedtoperformreasonablywell.The
twopresentedmeasurementresultsaboveoriginatefromtheSouthernOceanclosetoAntarctica.
Here,oneisconcernedwithrelativelysmallopticalthicknessesofallincludedabsorbersingeneral.
IncomparisontothebackgroundregioninthePacific,e.g.,theretrievedNO2amountishardly
differentinthisremotesiteandalsotheopticaldepthsoftheothereffectsaresmall.Thisnaturally
causesthenoiseleveltobemoreprominentthanincasesofstrongabsorptionsignals.Moving
overtoacompletelydifferentareaonEarth,e.g.tothehighlyindustrialisedAsiancities,the
fitresultsforNO2lookquitedifferent.OneexampleisdisplayedinFig.2.6(a)foratotalNO2
slantcolumnof2.64±0.05×1016molec/cm2(rms=1.6×10−4).Thisretrievalistakenfroma
measurementonMarch1st,2006,at40.75◦N124.25◦E,andthespectralfitexhibitsonlyasmall
relativeerrordemonstratingagoodperformanceoftheNO2fitting.Figure2.6(b)showsglobal
resultsfortroposphericNO2averagedoverthreemonths(Mar-May2006).ThestratosphericNO2
hasbeeneliminatedfromthetotalcolumnbysubtractingtheNO2amountfromareferencesector
(180◦-200◦East,wheretroposphericamountsarenegligible).Thisapproximationassumesthe

63

2Developingtheretrievalofiodinemonoxidefromsatellite

(a)

(b)

Figure2.5:TwoexamplefitresultsofthestandardIOfitincludinggraphsforthedifferential
Ringabsorptioneffectand(top),IO.theTheremainingtracegasfitsresidualshow(btwoottom)curvandesieacnhb-etwtheeenactualthefittracefromgasthefitsformeasuremenNO2,thet
stillcontainingtheresidualstructure(reddottedlines)andtherespectivereferencecrosssection
scaledwiththeretrievedslantcolumn◦amount◦(bluesolidline).Bothmeasurementsb13elongtoorbit2
rms=120051001_110..7×10−4(a),wandas(b)takenatat64.770.7◦S,S,48.753.7◦WWwithwithananIOIOamounamounttooff1.32.1±±0.40.5××101013molecmolec//cmcm2,,
rms=1.5×10−4.

64

2.2TheDOASretrievalofiodinemonoxide

stratosphericNO2amounttobeconstantwithlongitude,deviationsfromthisassumptionsoccur
butarenotrelevantinthiscase.
NodetailedanalysisoftheNO2resultsshallbegiven,butitshallberecognisedthattheimportant
featuresofNO2likethehighlyindustrialisedandpollutedareasofEurope,AmericaandAsiaare
capturedbytheretrieval.ThespatialdistributionagreeswellwithreportedNO2retrievals(Richter
etal.,2005),whiletheabsolutevalueslieintherightorderofmagnitude,butaresomewhatsmaller
inurbanareas.Thisismainlyduetotheuseofacrosssectionatcoldtemperatures(223K),
whichissuitablefortheIOretrievalbutgivestoosmallcolumnamountsforNO2,becausethe
absorptioncrosssectionisstrongerthanatwarmertemperatures(e.g.,293K).Additionally,onlya
roughconsiderationoftheAMFhasbeenperformedwithoutoptimisingtheprocedureforprecise
NO2verticalcolumns,butthemainpointhereliesonfitqualityandnotontheabsoluteamount
oftheNO2verticalcolumn.TheNO2analysisshowsthatthestandardIOfitperformswellalso
gases.traceotherfor(b)(a)

(b)

Figure2.6:(a)AtypicalNO2fitresult(rms=1.6×10−4)wherethescaledlaboratorycross
sectionandthemeasurementagreeverywell.(b)GlobaldistributionofNO2forthreemonths
(March-May)in2006,whereindustrialisedareasandalsosinglecitiescanbedistinguished.The
absoluteNO2amountsaretoosmallduetothechosencrosssectiontemperatureandtherough
considerationoftheAMF,butthisisnotrelevanthere,asthisscalinghasnoinfluenceonthe
qualityoftheIOretrieval.

ResultsfortheRingeffectintheIOstandardfit
TheresultsfortheRingeffectfromtheIOretrievalareacrucialpartofthisstudy.Inseveral
retrievalsotherthantheIOstandardfit,thespectralcorrelationbetweentheIOabsorptioncross
sectionandtheRingeffectspectrumleadstoerroneousslantcolumnsandfitfactorsforIOand
theRingeffect,respectively.Thechoiceofunfavourablefitparametersthenleadstoanirregular
spatialcorrelationbetweentheIOandRingpatternsinseverallocationsonEarth.Thereisno
physicalreasonwhyatmosphericIOamountsshouldbewellcorrelatedtotheRingeffect,although
insomecasescertaincausalitiesmayleadtosimilarbehaviourinbotheffects.Thedetectionofa
generallystrongspatialcorrelationbetweentheresultswouldpointoutaninterferenceduringthe
retrievalprocedure.Consequently,theRingeffectresultsserveasanadditionalcriterionforthe
judgementofthefitquality:theRingfitfactorsshouldyieldmeaningfulvaluesandalsothespatial

65

2Developingtheretrievalofiodinemonoxidefromsatellite

RingpatternshouldnotbedetectedintheIOresultinthesameform.

Figure2.7:GlobalmapoftheRing
moneffectthsfitfromfactoravSeptemeragedberotvoerNothevembthreeer
oc2005.currenceNegativofefitrotationalfactorsRamandenotescat-the
foundteringifthewhileRRSapisositivsmallerefitthanfactorintheis
kgroundbacectrum.sp

ForthestandardIOfitparameterset,anexampleoftheretrievedfitfactorsfortheRingeffect
isplottedinFig.2.7.Thedataisaveragedoveraperiodofthreemonths,SeptembertoNovember
2005.TheRingeffectisexpectedtobestrongoverlowgroundelevation,whilecloudsandhigh
surfaceelevationssuchasmountainrangesandplateausleadtoasmallerRingeffect.Additionally,
thefitfactordependsontheSZAwithlargerRingeffectforlargerSZA.Anadditionalsmalleffect
isinducedbythesurfacereflectancewithdecreasingRingeffectforbrightersurfaces,becausethe
Ramanscatteredradiationisrelativelyreducedincomparisontothereflectionattheground.These
dependenciesoriginatefromtheindividualphotonpathlengthsthroughtheatmosphere,especially
fromthepathlengthinthelowestatmosphericlayers,astheprobabilityforRamanscattering
becomeshigherforhigherairdensity.Theinfillingincreases,foralargeratioofRamantoRayleigh
scattering.Duetothedefinitionoftheeffectivepseudo-absorberspectrumfortheRingstructure(Sec.1.6.3),
astrongRingeffectcorrespondstonegativefitfactors.Thefillinginofabsorptionlinesappears
likeanemissionprocess.AsanEarthshinebackgroundspectrumisused,theresultsarerelative
totherespectivelocationovertheSouthPacific.PositivefitfactorsthereforemeanaRingeffect
whichisweakerthanatthatlocation.
Regardingtheseconsiderations,thefitfactorsoftheRingeffectinFig.2.7showmeaningfulval-
ues.Thefitfactortendstostrongernegativevalues(green/blue)forhigherlatitudesduetothe
increasingsolarzenithangle.Inaddition,severalmountainrangesandregionswithhighsurface
elevationexhibitpositive(red)valuesasexpected.Forexample,theHimalayaandtypicallycloudy
regionsinthetropicscanbedistinguished.EventheAlpscanbeidentified,especiallyinlonger
termaverages,butalreadyhereasyellowareasurroundedbygreencolour.
Inconclusion,theretrievalyieldssensibleandexplicableresultsfortheRingfitfactorsandis
notdisturbedbyretrievalerrorsorsystematiccorrelationwiththeIOamounts.Ascanbeseen
laterintheglobalIOmaps(cp.Fig2.12),therearenoprominentfeaturesintheIOresultwhich
wouldcorrelateonlywiththeRingeffect.Theseobservationsgivefurtherreassurancetothequality
ofthestandardIOretrieval.

66

2.3AirmassfactorconsiderationsfortheIOretrieval

2.3AirmassfactorconsiderationsfortheIOretrieval

AsdescribedinSec.1.8,theAMFtransformstheslantcolumnintoaverticalcolumn.Inorderto
computetheverticalcolumnfromthesatelliteslantcolumnresults,severalassumptionsareneces-
sary.Inthepresentcase,somefactorsarenotwellknownandthereforetheregulartransformation
ofslantintoverticalcolumnswouldintroducelargeuncertainties.Therefore,theslantcolumnsare
retainedandastandardconversionintoverticalcolumnsneedstowaituntiltheopenquestions
havebeenanswered.Nevertheless,thecomputationofAMFsisofinterest.Foragivenvertical
column,thepresentAMFdescribesthesensitivityofthesatellitemeasurementforthetracegas
detection.Therefore,someconsiderationsontypicalAMFsforsomegivencasesarenecessaryfora
betterinterpretationoftheIOresults.
Ingeneral,alargeslantcolumnamountmaybeenhanceddueto(atleast)twodifferentcauses.
Eitherthetracegasamountmightbelargeor,alternatively,theAMFmaybeenhanced.The
firstcaseisobviousandthesecondcaseevolvesasaconsequenceofsurroundingconditions.The
lightpaththroughatracegaslayerisaffectedby,e.g.,thesolarzenithangle,thegroundspectral
reflectance,theaerosolloadintheatmosphere,cloudsandthetracegasaltitudeprofile.These
aspectsneedtobetakenintoaccountifabsolutevaluesareinterpreted,andsometimesalsoif
qualitativecomparisonsaremade.
BlockAMFshavebeencalculatedforapureRayleighatmosphereandseveraltypicalconditions.
Foriodinespecies,themid-latitudemarineboundarylayeraswellasPolarRegions,whichare
typicallyicecovered,areofspecialinterest.TheSZAandthealbedohavebeenvariedaswellas
theinfluenceofalowandstronglyreflectingcloudhasbeenbeeninvestigated.

SZAtheofinfluenceTheForAntarcticlocationslikeHalleyStation(cp.Sec.3.2.2),themimimumSZAliesaround55◦and
maximumSZAusedinthedataproductis84◦.Figure2.8showstheblockAMFforSZAsof55◦,
70◦and80◦ataltitudelevelsbetweentheEarthsurfaceand10kmforanalbedoof90%.For
IObeingsituatedatlowaltitudes,probablyeveninthelowestpartoftheboundarylayer,the
influenceoftheSZAonthenear-surfaceblockAMFisontheorderof10%.Intheupperpart
ofthetropospheretheAMFsdifferstrongly(byafactorof2).However,towardstheground,the
blockAMFsconvergetovaluesbetween3.8and4.4.

Theinfluenceofthealbedo
Forasolarzenithangleof70◦,theinfluenceofachangingalbedoisdemonstratedinFig.2.9.
Especiallyforlowaltitudes,theeffectishuge.Increasingthealbedofromatypicalvalueof5%
(e.g.overtheocean)to90%(forbrightsnoworice)changestheblockAMFdirectlyattheEarth’s
surfacebyafactorof6from0.7toabout4.1.ThetotalAMFforanIOprofileinthelowest1km,
differsbyafactorof4.
ThisobservationisimportantforIOretrievalsastheamountsareoftenclosetothedetectionlimit,
whichisstronglyinfluencedbytheAMF.Thedeterminationofthedetectionlimitisthetopicof
thenextsection.ConsideringagivenIOamount,itmightbebelowdetectionlimitoveradark

67

2Developingtheretrievalofiodinemonoxidefromsatellite

surface,butwellobservableoverabrightunderground.

Figure2.8:Comparisonofblockair
massfactorsatdifferentsolarzenith
TRANSCIAwithcalculatedanglesVersion2.0(Rozanovetal.,2005b)for
425nmandanalbedoof90%.Thein-
fluenceoftheSZAislargeintheupper
tropospherebutrathersmalltowards
theEarth’ssurface.ForIOamounts
inthelowestlayers,theSZAtherefore
hasnotsuchabigrelevance.

InfluencefromtheIOaltitudeprofile
AsoceanicareasbelongtotheinterestingregionsforIO,itisalsoimportanttonotethestrong
altitudedependenceoftheblockAMFforthiscase(e.g.Fig.2.9,bluecurve).IftheblockAMF
changesalotwithheight,theassumedtracegasprofilewillhaveaneffectonthecalculationofthe
totalintegratedAMF.ForaSZAof70◦andanalbedoof90%(redcurve),theblockAMFisnearly
constantwithheight.ForallotherexamplesofvaryingSZAandvaryingalbedo,thisisnotthe
caseandthetracegasprofilewillhaveastronginfluence.Dependingontheindividualsituation,
theverticalcolumnmaychangebyafactorof∼3,betweenanIOprofileconfinedtotheboundary
layerordistributedoverthetroposphere.

ImpactofcloudsontheAMF
TheimpactofcloudsontheAMFandespeciallyontheblockAMFprofilecanbequitelarge.
Theexactcalculationalsodependsstronglyoncloudparameterslikeliquidwatercontent,cloud
structureanddropletsizes.NodetailedanalysisofcloudinfluencesontheAMFshallbegiven
here,butonecentralaspectshallbepointedout.Thetypicallightpathenhancementwithinthe

68

Figure2.9:Comparisonofblockair
massfactorsfordifferentalbedositua-
tionscalculatedwithSCIATRANVer-
sion2.0(Rozanovetal.,2005b)for
425nmandaSZAof70◦.Thestrong
influenceofthegroundreflectanceon
theblockAMFclosetothesurface,
andconsequentlyalsoonthesensitiv-
itytowardsIOdetectionisvisible-
theblockAMFclosetothesurfacein-
creasesfrom0.7at5%to4.1at90%.

2.3AirmassfactorconsiderationsfortheIOretrieval

cloudtopisalsoneglectedhere,onlytheeffectonthepathabovethecloudshallbeconsidered.
ThreeblockAMFsarecomparedforthispurposeinFig.2.10.Allcurvesarecalculatedforequal
conditions,exceptforthealbedoandthepresenceofacloud.TheSZAis55◦inallthreecases.
Twocurvesreachdowntothegroundandarecalculatedforanalbedoof5%(green)and90%(light
blue).Thestrongbluecurveiscalculatedforabright100%reflectingcloud(novisibilitybelowthe
cloud)at1km.Thecomparisonshowsthatthedifferencebetweenthe90%albedocaseandthe
cloudcoveredsituationissmallabove1km.ThesensitivitytowardsIOintheupperlayerswould
bethesameandIOamountsbelowthecloudwouldsimplybeignored.Thesituationisdifferent
overadarksurface.TheblockAMFabove1kmchangesstronglybetweenthe5%reflectingground
andthecloudyscene.TheAMFabove1kmismuchlargerabovethecloud.Consideringsatellite
measurements,thismeansthatapartlycloudypixelwillcarrymostinformationfromthecloud
coveredpartdominatingtheinformationcontentcomingfromthecloud-freeportion.Forthe90%
albedocase,IOamountsinthecloud-freepartwouldstandalargerchancetobeobservedinspite
ofsomemaskinginthecloudyregion.

Figure2.10:Comparisonofblockair
massfactorsforthreedifferentcases,
analbedoof5%,analbedoof90%and
forthecaseofa100%reflectingcloud
at1kmaltitude.Whilethedifference
above1kmisverysmallbetweenthe
cloudycaseandthehighlyreflecting
surfacecase,thechangeislargeover
adarksurfacewhenacloudispresent.

rySummaThemainoutcomeoftheanalysisoftheAMFarethefollowingpoints:
•TheoverallAMFforIOcanvarystronglydependingonconditions.
•ThesolarzenithanglehasonlyasmallinfluenceontheAMFfortracegasesclosetothe
ground,aslongastheSZAdoesnotexceed80◦.TheSZAinfluenceismuchlargerathigher
altitudes,andisthereforeprobablynothighlyrelevantforIO.
•Thealbedo,incontrast,showsastronginfluenceespeciallyclosetotheground,wherethe
blockAMFmayvarybyafactorof6betweenextremeconditions.TheintegratedAMFfor
IOwilltypicallyvarybetween1andabout4.
•Stronglyreflectingclouds,whichhidetheatmospherebelowfromtheviewofthesatellite,
provokelargestchangesifthesurfacebelowisdark(albedoof5%),whileforabrightsurface
themaineffectisthesimpledisregardofthelowlayers.

69

2Developingtheretrievalofiodinemonoxidefromsatellite

•DuetotheuncertaintiesintroducedbytheunknownverticalprofileofIO,theslantcolumns
willnotbeconvertedtoverticalcolumnsinthestandardIOproduct.However,theAMF
considerationsabovearevaluableinformationneededfortheinterpretationofIOamounts,
temporal/spatialvariancesanddetectionlimitsinindividualcases.

2.4DetectionlimitforIO
Thedetectionlimitofameasurementstatesthesmallestamountofatracegas,whichcanstillbe
identifiedbytherespectivemethod.Thisisanimportantvaluetobeconsideredforthejudgement
ifameasurementtaskispromising.
InordertoestimatethedetectionlimitofSciamachyfortheidentificationofIOslantcolumns,
severalfactorsdeterminedbytheinstrument,themeasurementprocedureandalgorithmaswellas
thephysicalprocessesneedtobeconsidered.Arelevantparameterinthisrespectisthesignal-
to-noiseratio(S/N)ofthemeasurement.TheminimalopticaldepthODminstilldetectableby
theinstrumentisdeterminedprimarilybytheinverseofthisvalue,theratioofthenoisew.r.tthe
signal(N/S),whichwillbeshownbelow.

ratiosignal-to-noiseTheThesignalSisgivenbythenumberofelectronsgeneratedbyincomingphotonswithphotonflux
Fγ(inphotons/s/pixel)atthedetectorduringthemeasurementintegrationtimet(ins)andwith
quantumefficiencyqe(electrons/photon):

S=qe·t·Fγ
Thesignalstrengththereforedependsonparameterslikethespectralregion,thesurfacespec-
tralreflectanceandthesolarzenithangleincombinationwiththerespectiveintegrationtime.
ThenoiseamountNisthevalueforthetotalnoiseperdetectorpixelarisinginthemeasurement
processandcontainsalldisturbingeffects,mainlyfromthreedifferentprocessesgeneratingaddi-
tionalsignalsonthedetector.ThesenoisesignalsarethestatisticalphotonnoiseNi,theread-out
noiseNrandthenoiseofthedarkcurrentNd.Asthephysicalemissionprocessisofstatistical
nature,thephotonsignalisdeterminedbyPoissonstatisticsandisthereforedescribedbyaPoisson
distribution,whichconvergestoanormaldistributionforlargephotonnumbers.Consequently,the
statisticalnoiseNi(shotnoise)isgivenbythesquarerootofthesignalNi=√S.Theread-out
noiseisgeneratedbytheelectronicsduringtheread-outprocedureandisindependentofintegration
timeandsignalstrength.Thedarkcurrentwhichispresentalsoifnoradiationmeetsthedetector
hasastatisticalnoisesignalaswell.ThemagnitudeofNdincreaseswithintegrationtimeofthe
measurementandstronglydependsonthedetectortemperature.Itiseffectivelysuppressedby
detectorcooling.Intotal,thesignal-to-noisesignalisgivenby:
SS=NNi2+Nr2+Nd2

70

2.4DetectionlimitforIO

Inthepresentcase,theread-outnoiseissmalland,especiallywiththegivenlargephotonflux
inthevisiblespectralregionanda√wellcooleddetector,√thedarkcurrentnoiseismostlynegligible
incomparisontothesignalnoiseS.Thisyields:NS≈S.
depthopticaldetectableminimumTheTheopticaldepthODbetweentheactualintensitymeasurementI=S±Nandthebackground
measurementI0iscalculatedinthefollowingway:
OD=lnII0=ln(I0)−ln(S±N)=ln(I0)−ln(S)−ln(1±SN)
N≈ln(I0)−ln(S)∓S
ODrealerrorODThefirstterminthissumistheundisturbedopticaldepthODreal,whilethesecondistheerror
intheopticaldepth.TheapproximationmadeinthelaststepholdsifNS,whichisvalid
forasufficientlyhighmeasurementintensity.IfI0ischosenfromanEarthshinemeasurement,the
noisecanbereducedbyaveragingandinthatcasebeneglectedincomparisontothenoiseofeach
individualmeasurementI.Otherwise,thenoiseinthebackgroundmeasurementwill,ofcourse,
furtherincreaseODerror.DemandingastrongerODsignalthantheerrorODforreliabledetection,
leadstotherequirementoftheminimumdetectableopticaldepthODmin:
ODmin≈N≈√1
SSForthewavelengthregionunderconsiderationhere,thetheoreticallyexpectedS/N-ratioisonthe
orderof5000foranalbedoof90%and2000for5%albedo(Noëletal.,1998).Theerrororrootmean
squareoftheopticaldepth(ODrms)isthenontheorderof10−4dependingonconditions.Typical
experimentalODrmsvaluesfromtheIOretrieval(cp.Sec.2.2.2)liearound2×10−4(between
1-3×10−4).
Thecomparisonbetweentheexperimentaldetectionlimitsandthetheoreticallyestimated
numbersallowstheconclusionthattheretrievalisclosetooptimalperformance.Thetheoretical
limitisonlyslightlysmallerthantheexperimentallyidentifiedvalues.Muchstrongerreductionof
theresidualrmsvalueismostprobablynotpossible.Someremaininginfluenceontheresidualis
alsocausedbysystematiceffectsandishenceindependentofaveraging.

Consequencesfortheslantcolumndetectionlimit
Foranidealmeasurement,theslantcolumndetectionlimit(SClim)isgivenbytheratioofthe
residualopticaldepthODrmsandthemaximumdifferentialabsorptioncrosssectionvalueofthe
respectivetracegasσmax.

SClim=ODrms
σmaxThisanalysisassumesthatatracegasisdetectableifitsabsorptionopticaldepthbecomes

71

2Developingtheretrievalofiodinemonoxidefromsatellite

largerthantheresidualrmsvalue.ForIO(σmax=2.8×10−17cm2/molec)andaresidualrmsof
2×10−4(typicalfor90%albedo),thedetectionlimitisgivenbySClim=7×1012molec/cm2fora
singlemeasurement.Thislimitcanbefurtherreducedbyaveraging,intimeorspace,providedthe
sourceoferrorsisrandomandsystematicerrorshavebeenaccountedfor.ThetheoreticalS/N-ratio
at5%albedocorrespondstoanrmsvalueofabout5×10−4andanIOslantcolumndetection
limitof2×1013molec/cm2forasinglemeasurement.Forthespatiallyaveragedgroundsceneof
60×120km2,usedinthisstudy,thetheoreticalSClimitisreducedbyafactorof2.
Forexperimentalresults(i.e.alsoforagroundscenesizeofatleast60×120km2),with
theabovementionedODrmsvaluesaround2×10−4,thetypicalIOslantcolumndetectionlimitis
7×1012molec/cm2.Thislimitisthenfurtherdecreasedbytemporalaveraginginasubsequent
step.FormonthlyaveragesandforaSciamachyoverpasseverysixdays,theSCdetectionlimit
isreducedtoaround3×1012molec/cm2.Thisreductionobviouslydependsonthenumberof
overpassesandtheaveragingtimeperiod.

Theconversiontoverticalcolumnsandmixingratios
Thedetectionlimitisusefulforanestimationifthesatelliteobservationsarecapableofidentifying
expectedtracegasamounts.ManymeasurementsofIOhavespecifiedthegroundvolumemixing
ratioinsteadofacolumnamount.Inordertojudgeiftheseamountsshouldbedetectableby
Sciamachy,thedetectionlimitontheslantcolumnneedstobeconvertedtoadetectionlimitfor
thesurfacemixingratio.Naturally,thisconversionwillintroducetheusualuncertaintiesdiscussed
previously.AssumptionsontheappropriateAMF,andinparticularonthealtitudeprofileofIO
arenecessary.Thealtitudeprofileenterstheconversiontwice,onceintheAMFcalculationand
anothertimeintheconversionfromtheverticalcolumntothemixingratio.Accordingtocurrent
knowledge(cp.Sec.1.5),IOisprimarilyatropospherictracegas(Friessetal.,2001)withevidence
forconfinementtothelowerpartoftheboundarylayerSaiz-Lopezetal.(2007c).AssumingtheIO
tobesituatedinthelowest1kmorinthelowest100misconsequentlyreasonable.

culatedFigure2.11:withBlocSciamakairchymassVersionfactorscal-2.0
Rozanovetal.(2005b)for425nm,and
talbwoedo,relev70an◦tSZAexample(tloypicalcations,forAnred:tarctic90%
coast),blue:5%albedo,55◦SZA(typical
forIreland).ThesensitivityforIOdetec-
tionisseveraltimeshigherforAntarctic
conditions.

FollowingtheAMFdiscussionintheprevioussection,twotypicalcasesarechosenhereas
examples.ForatypicalsituationattheIrishcoastwith5%surfacereflectanceand55◦SZAthe

72

2.4DetectionlimitforIO

totalAMFthenamountstoapproximately1.0,incontrastto4.2foratypicalcaseforAntarcticaat
90%albedoand70◦SZA.ThesevaluesholdforanIOboxprofilewithinthelowest1km(withlinear
concentrationdecreasetozerouptothenextaltitudegridpointat1.2km).Figure2.11showsthe
blockAMFforthetwodifferentsettings,similartothegeneralcasesdiscussedinSec.2.3.The
largersensitivityforthedetectionofIOundertypicalAntarcticconditionsisagainvisible.Forthe
Antarcticcase,theAMFishardlyprofiledependent,whileforthemid-latitudecoastalsituation
theAMFreducesto0.8iftheIOisassumedtobewellmixedonlyinthelowest100m.

Theoreticaldetectionlimitsforindividualmeasurements
coastIrishtarcticAnSC(molec/cm2)7×10127×10122.0×10132.0×1013
boxprofileheight1km100m1km100m
0.81.04.24.2AMFVC(molec/cm2)1.7×10121.7×10122.0×10132.5×1013
VMR0.7ppt7ppt8ppt100ppt
Experimentaldetectionlimitsformeasurementsaveragedover60×120km2
coastIrishtarcticAnSC(molec/cm2)7×10127×10127×10127×1012
boxprofileheight1km100m1km100m
0.81.04.24.2AMFVC(molec/cm2)1.7×10121.7×10127×10128.8×1012
VMR0.7ppt7ppt2.8ppt35ppt
Table2.4:DetectionlimitsforSciamachyobservationsofIOforsometypicalconditions.The
detectionlimitsfortheslantcolumns(SC)resultfromtheoreticalconsiderations(toppanel)and
experimentalevidenceupperlimits(bottompanel).Conversionofvaluesyieldtheverticalcol-
umn(VC)andthevolumemixingratio(VMR)independenceoftwoassumedboxprofileheights
(withhomogeneousmixingbelowthisheight)andthecalculatedAMF(usingtheSCIATRAN2.0
code).Formonthlyaverages,typicallyadditionalreductionbyafactorofaround2canbeachieved
dependingonthenumberofoverpasses.

Usingtheslantcolumndetectionlimitsdeterminedfromtheoreticalconsiderationsaswellas
fromtheexperimentalfindings,thedetectionlimitsfortheverticalcolumnsandvolumemixing
ratiosarecalculated.Forseveralrelevantcasesandassumptionsdiscussedabove,theresultsare
summarisedinTab.2.4.Stronglyinfluencedbychosenconditions,VMRdetectionlimitsarehighly
variableandmayliebetween0.7pptand35ppt.

limitsdetectionofDiscussionTheanalysisshowsthatIOamountsreportedbyground-basedmeasurements(Sec.1.5.5)arecloseto
theIOdetectionlimitsforasingleSciamachygroundscene.Oftenthedetectionlimitwillinhibit

73

2Developingtheretrievalofiodinemonoxidefromsatellite

theIOobservationfromsatellite.Insomecases,especiallyintheAntarctic,reportedamountsare
indeedsomewhathigherthanthedetectionlimit.Forthesecases,positivedetectionofIOfrom
satellitecanbeexpectedunderanadditionalcondition.ThespatialextentoftheenhancedIO
amountsneedstobeontheorderofasatelliteground-pixel.Thisisanimportantpoint,because
theavailableground-basedandballoonmeasurementsmadesofarareunabletodeterminethe
spatialextentofIOreleaseanditisunknownhowlocalisedenhancedIOoccurs.IOreleaserelated
toalgaeemissionsmaybeconfinedtoverylocalspots.Inthatcase,themixingratioswouldneed
tobemuchlargertoenableSciamachytodetecttheIOabsorption.

2.5PrecisionandaccuracyoftheIOretrieval
Everymeasurementresultisinherentlyaffectedbysomeuncertainties.Theseeffectsarecausedby
statisticalaswellasbysystematiceffects.Whilestatisticalerrorsmaybereducedthroughaver-
aging,systematicinaccuraciesremainalsoaftertheaveragingprocedure.Ingeneral,thestatistical
errorsofanexperimentaffectitsprecision,whilesystematicdeviationsdeterminetheaccuracyof
t.measuremena

rserroStatisticalThediscussionofthedetectionlimitintheprevioussectionalreadyconsidersthestatisticaleffects.
Consequently,thedeterminationofthedetectionlimitconcurrentlyyieldstheprecisionofanin-
dividualmeasurement.Thevaluescomputedfortheslantcolumndetectionlimitcanbedirectly
transferredtothecolumnprecisionasthenoiseintheopticaldepthconsideredabove(ODerror)
leadstostatisticallyfluctuatingslantcolumnresults.TheIOslantcolumnresultsfromsatellite
measurementspresentedinthefollowingchaptersaretypicallyaveragedoveratleast3monthsand
thereforeexhibitbetterprecision.Thismayimprovetheprecisionbyafactorof4ascompared
totheexperimentalerrorvalueswhichareusedforthedetectionlimitcalculationfromindividual
spectraabove.
Fromthefindingsintheprevioussectionthattheexperimentallyidentifiedresidualsareat
mosttwotimeslargerthantheoreticallypossible,itcanbefollowedthatamajorpart(atleast
50%)oftheremainingresidualisusuallycausedbystatisticalfluctuations.Thestatisticalerrorin
seasonalaveragesofIOslantcolumnsisthencalculatedtobeatleast1×1012molec/cm2.

Inaccuraciesduetospectralcorrelationsandscalingerrors
Theaccuracyisbasicallydeterminedbythreedifferenttypesofsystematiceffects.Thefirsttype
causesinterferencesduringtheretrievalprocess.Atypicalexampleisthespectralcorrelation
betweentwoormoretracegasesconsideredintheretrieval.Inaccuraciesinthecrosssection
spectra,approximationsinthecalculationandconsiderationoftheRingeffectandstraylightas
wellasunidentifiedspectralstructuresandleftoverinstrumentaleffects(suchasthememoryeffect
orsimilar)alsobelongtothiscategory.Theconsequenceareresidualfeatures,whichincreasethe
possibleerrorintheretrievedtracegasamounts.
Thispartofthesystematicerrorsisminimisedbythespecificchoiceofretrievalparameters.Largest

74

2.5PrecisionandaccuracyoftheIOretrieval

systematicerrorsourcesarethuslargelyexcludedbytheconsistencycheckspresentedinSec.2.2,
butsomeparticularinaccuraciesmightremain.InSection3.1,someirregular,negativeslantcolumn
valueswillbepointedout.Thesevaluesareontheorderof3×1012molec/cm2inaverageddata,
sosomesystematicinfluenceonthisorderofmagnitudefromspectralcorrelationsispossible,but
canbelargelypreventedbyconsistencychecks.
Asecondtypeofinfluencesarescalingerrors,whicharenotattributedtodisturbingeffects
duringtheretrievalprocess.Theseerrorswouldevenoccurunderperfectmeasurementandretrieval
conditions,i.e.,especiallyalsoforsimulatedmeasurementswithoutnoise.Oneprominentexample
ofthisclassoferrorsisthetemperaturechoiceforthecrosssectionreference.Allabsorption
crosssectionsdependonthetemperatureoftheabsorbingmolecules.Duetothermalenergyand
theresultingoccupationprobabilitiesofenergylevels,linestrengthstypicallydecreasewithrising
temperature.TheappliedIOcrosssectioninthisstudyhasbeenmeasuredinthelaboratoryatroom
temperature,298K(GómezMartínetal.,2005;Spietzetal.,2005).TemperaturesinAntarctica,
e.g.,arecolderandthereforetheabsorptionstrengthpermoleculeisstronger.Asaconsequence,
theretrievedIOamountwillbesystematicallytoohighbysomepercentage.IOspectraatother
temperaturesarenotavailable,butestimationsusingBrOasmeasureforcomparisonarepossible,
forwhichcrosssectionshavebeendeterminedbyFleischmannetal.(2004)atvarioustemperatures.
Adjustingthedifferentialabsorptionlinemagnitudemeasuredat298Ktotheoneat243K,e.g.,
requiresascalingby17%.Dependingonregionandtemperatures,theeffectmaybedifferently
strong,butgenerallyrangesbetween0and20%.
Additionally,athirdcategoryofindividualinfluencescanbedistinguished,withmostlyvarying
impactandtypicallyleadingtooffsetsontheretrievedamounts.
Thepresenceofcloudsmayintroducesuchadditionalerrors.TheimpactofcloudsontheAMF
hasbeenmentionedinSec.2.3andtheeffectofcloudsontheretrievedIOslantcolumnmapswill
bediscussedseparatelyinSec.2.7.Inprinciple,cloudsmayobstructtheIOcolumnpartlyfrom
thesatellite’sview,sothattheretrievedIOcolumnsareexpectedtobetoolowinthepresenceof
cloudsabovetheIOlayer.IncaseofoverlappingvolumesofcloudsandIOpresence,theinfluence
definite.lessisAnadditiveslantcolumnoffsetmayresultfromthechoiceofthebackgroundEarthshinespectrum.
IncasesomeIOispresentatthechosenlocation,thiswillbesubtractedfromtheotherregions.
However,thebackgroundreferencelocationhasbeenchosensuchthatnoIOisrecognisedthere.
Dependingonhowthebackgroundspectrumisconsideredandcalculated,theresultingIOcolumn
canvaryslightly.Calculationsshow,thattheimpactisontheorderofbelow1×1012molec/cm2.
However,thisisthesameforallregionsonEarthanddoesnotinfluenceanindividualmeasurement
orlocation.Spatialcomparisonsremainentirelyunaffectedbythissourceofuncertainty.

Summaryoferrorestimates
Statisticalerrors:±1×1012molec/cm2.
Systematicerrors:±[20%+1×1012molec/cm2+(3×1012molec/cm2)].
Thelasttermof(3×1012molec/cm2)issetinparenthesesasthiscontributioncanbelargelysup-
pressedandavoidedbyconsistencytestsandqualityinspectionoftheoverallfitresult.Including

75

2Developingtheretrievalofiodinemonoxidefromsatellite

theadditionalterm,theoverallerrorcouldexceed5×1012molec/cm2.Typicaloverallerrors,how-
ever,lieontheorderof±2-3×1012molec/cm2.Onthecolourcodedmapsofsatellitedatatheerror
sourcesandresultingerrorbarsarenotrepresentable,sotheabovediscussionneedstobekeptin
mind.

2.6ExampleresultsofglobalIOcolumns

Inthenextsections,spatiallyresolvedmapsofIOareusedfordiscussionsandcomparisonsofthe
qualityandconsistencyofthefinalIOproduct.Asapointofcomparison,afirstmapofthestandard
IOcolumnsdiscussedaboveisplacedinFig.2.12,alsoaspublishedinSchönhardtetal.(2008).
Inthefollowingdiscussions,influencesontheIOresultsandpotentialirregularitiesintheproduct
canthenbeassignedandhighlightedaccordingly.Noscientificinterpretationanddiscussionofthe
observedIOvaluesthemselvesorthespatialandtemporaldistributionisgivenhere.Thisissubject
3.ChapterofInFig.2.12,theresultsoftheIOslantcolumnsareshownonaglobalmapoftheEarth.Forthe
resultsofthestandardIOretrieval,thesamecolourscale(givenontherightinFig.2.12)isused
throughoutthisstudy.Thechosencolourscalecombinesvaluesbelow3×1012molec/cm2intolarger
intervalsandisfollowedbysmallerandlinearstepsabove.Themapshowsanaverageoverthree
monthsoftheyear2005,fromSeptembertoNovember.ThistimeperiodofSouthernHemispheric
Springwillbecomeimportantinthescientificanalysisanddiscussion(cp.Sec.3.2).Thevalues
enteringthesefinalmapsofIOamountsobeyaqualitycriterionlimitingtheopticaldepthrms
tovaluesbetterthanorequalto4·10−4.(Thiscausesresultstobediscarded,ifthefitresidual
becomeslargerthantwicethatspecificnoiselevelwhichwouldleadtothetypicalexperimental
detectionlimitof7·1012molec/cm2fortheIOslantcolumn.)

screeningCloud2.7

Whilecloudsarenoprobleminsomeotherwavelengthrangessuchasthemicrowave,visiblera-
diationexperiencesscatteringandreflectionatclouds,sothatmeasurementsareinsomecases
considerablyaffected.Cloudsinthefieldofviewcancauseapositiveornegativebiasforthere-
trievedtracegasamountdependingonthealtitudeprofileofthetracegasandoncloudproperties,
ormightintroduceothersystematicproblems.Inprinciple,itisdesirabletoapplyacloudscreening
algorithmtothesatellitedata,whichfiltersoutsceneswherecloudshaveaconsiderableimpact.
Nevertheless,nocloudcorrectionschemewasroutinelyappliedtothestandardIOproductfor
specificreasons.OnespecialinterestistheinvestigationofIOinthehighlatituderegions.Some
reflectivitypropertiesoficeandsnowareverysimilartothoseofcloudssothatareliableseparation
ofcloudsfromiceandsnowisnotroutinelyavailableyet.Ofcourse,itispossibletoapplysome
cloudcriterionintheotherregionsonEarth,butnotforaconsistentglobaldataset.Whilethe
advantageofapplyingacloudscreeningisthereductionofpossiblesystematicerrors,anessential
disadvantage,whichisespeciallypresentforretrievalsofsmallopticaldepths,isthelossofdata
andthereforeareducedbenefitfromtheaveragingprocedure.Itneedstobejudgedindividually,

76

screeningCloud2.7

Figure2.12:GlobalmapshowingslantcolumnresultsofthestandardIOfit.TheIOcolumns
areaveragedoverthetimeperiodofSeptembertoNovember2005andthecolourcodeappliedis
shownontheright.ScientificdiscussionoftheseresultsistopicofChapter3.

whichcloudscreeningsettingsareappropriateforwhichtask.
ThepotentialinfluenceofcloudsontheIOproductwasinvestigatedintwoteststudiespresented
inthefollowing.First,asimplecloudscreeningmethodwasusedforlowerlatituderegions,where
measurementsarenotinfluencedbysnowandice.Inasecondtest,anewandpromisingsurface
classificationmethod,whichhasrecentlybeendevelopedattheIUPBremen,willbeappliedto
observationsalsoabovehighlyreflectingregionsliketheAntarctic.Distinguishingicesurfacesfrom
brightcloudsissupposedtobepossiblewiththismethod,whichconsequentlyshowsahighpotential
forcloudscreeninginhighlatitudesforthefuture.

2.7.1Cloudscreeningwithanintensitycriterion

Concerningradiationinthevisiblewavelengthregion,onebasicpropertyofcloudsistheirhigh
reflectivity.Theoverallintensityofarecordedspectrumisinfluencedbytheamountofreflecting
cloudswithinthefieldofview.Fromthissimpleconsiderationcloudyscenesmayberoughly
ones.freecloudfromdistinguishedTherealityismorecomplex,asthereflectivitypropertiesofcloudscanbeextremelyvariable,high
andthickcloudsreflectandscatterphotonsdifferentlythanlowandthinclouds.Additionally,the
variationsofscatteringandreflectionpropertiesarenotclearlylinkedtotheeffectsonthetrace
gasproduct,i.e.,ahighlyreflectingclouddoesnotnecessarilyaffectthetracegasretrievalmost.
Theapplicationofanintensitycriterionneverthelessshowsifthemeasurementsaresystematically
affectedbycloudsandifartefactsareintroduced.
Asuitablelimitfortheoverallintensityneedstobechosen,fromwhichvalueonwardsascene
isdefinedastoocloudyandisdiscardedfromtheproduct.Thelimitneedstoliebetweenthe
lowestandhighestoccurringintensities,whereundercomparableconditionsthelowestintensity

77

2Developingtheretrievalofiodinemonoxidefromsatellite

correspondstoacompletelycloudfreeviewandthehighestoccursforaclosedcoverofbright
reflectingclouds.Thelowerthethresholdisset,theclearertheviewneedstobeforthemeasurement
topassthecriterion.Thereisnofixeddefinitionforthemagnitudeofthislimitanditwillinany
casedependonthewavelengthregioninwhichtheanalysisisperformed.
Inthepresentcase,avalueof1×105Counts/sprovidedasuitableintensitythreshold.Inorderto
applythiscriterion,twostepsneedtobetaken.Firstofall,therecordedintensityintherespective
fittingwindowisaveragedoverallspectraldetectorpixels.Secondly,theintensityvalueisdivided
bythecosineofthesolarzenithangle(validforLambertianreflection)toaccountfortherelative
darkeningpergroundareaofasceneatlowsun.
Inordertodeterminetheabovethreshold,Sciamachy’sPMDchannelsarehelpful.Fromtheir
differentwavelengthbands(cp.Sec.1.9.1),avisualcolourinthesenseofanRGB(Red-Green-Blue)
codecanbecalculated.Thisdataisreadilyavailableandcouldbeusedwithoutfurtherprocessing
withinthisstudy.WhenplottingthevisualimpressionfromthePMDvalues,cloudyscenesappear
brightwhiteoverdarklandoroceanregions.Figure2.13comparesSciamachymeasurements
overEuropeon1stSeptember,2005.Theleftmap(a)showsthevisualimpressiondeterminedby
thePMDdata.Cloudscanbedistinguishedfromwaterandlandinthisfigure,astheunderlying
surfaceisdark.Thisisnotpossibleforsnowandicecoveredregions.Incomparison,theright
map(b)displaystheintensityvaluerecordedbySciamachyinthestandardIOfittingwindow
accordingtothecolourscaleontheright.Lowintensitiesappeargreen,andhighervaluesare
plottedinorange/red.Thecolourchangetoredisintentionallysetatthechosenlimitof1×105
Counts/s.Theagreementbetweenredscenesinmap(b)andcloudyareasinmap(a)isreasonably
good,thecloudcoveredareas,e.g.,NorthwestofNorwayandoverFrancecanbedeterminedin
bothpictures.ItcanalsoberecognizedthatsomeSciamachypixelswhereonlyasmallfraction
isaffectedbyvisiblecloudswillnotbediscardedfromthefinalproduct.Theintensitycriterionis
thusnotabsolutelystrictandcertainpercentagesofcloudsarestillallowed.
(b)(a)

Figure2.13:Theleftmap(a)showsthevisualimpressionofthegroundsceneasdetermined
fromthedifferentPMDbands,whereclouds(appearingwhite)canbedistinguishedfromcloudfree
areas.Intherightmap(b),theSZAcorrectedintensityasaveragewithintheIOfittingwindow
isplottedforthesamedayandregion.Thehighestintensities(red)occurovercloudyareasand
wouldbediscardedintheintensitylimitedproduct.

AsanexampleofhowtheIOproductisaffectedbythisintensitycriterion,Fig.2.14compares

78

Cloud2.7screening

theIOproductwithoutanintensitylimit(a)totheresultafterapplicationofthethreshold(b).All
brightregionsaremissingintherightmap,especiallythesnowandicecoveredPolarRegionsand
Greenlandarecompletelymasked.Whenapplyingtheintensitycriterionmanymeasurementsare
discarded.Thisleadstoasubstantialincreaseinthefinalnoiseoftheproduct,whichisrecognised
bystronglyvaryingvaluesinlowlatitudes.ItcanbeseenthattheoverallIOvaluesdonotchange
systematicallyandnospecificartefactsareintroducedwhencloudyscenesarekeptinthefinaldata
product.Beforethisisnotchecked,onecannotknowifsomeenhancedIOvaluesmightbedueto
cloudrelatedinterferencesinthemeasurements.However,thisisnotthecasehere.
ThistypeofcloudscreeningisnotusefulingeneralfortheIOproduct,asalliceandsnowscenes
arecompletelylost.Themainresultofthiscomparisonisthatnolargesystematicerrorsoreven
artefactsareintroducedwhenomittingacloudscreeningprocedure.
(b)(a)

Figure2.14:ComparisonoftheIOfinalproductwithoutacloudscreeningprocedure(a)and
1after×105applicationCounts/sofinantheintensitresult.yThiscriterionway,(b)cloudywhichoscenesnlyleabutvesalsosnomeasuremenwandtsicewithcovineredtensitiesregionsbeloarew
removedandthespatialnoiseismarkedlyincreased.

2.7.2CloudscreeningusingthePMDbasedclassificationscheme
Onlyrecently,anewmethodhasbeendevelopedwhichclassifiestheunderlyingsurfaceandcloud
typeofeachSciamachymeasurement(Lotzetal.,2008).Thedifferentcategorieswhichare
consideredinthisclassificationschemearelistedinTable2.5.
ThemethodiscalledSPICS(Sciamachy-PMDbasedIdentificationandclassificationofClouds
andSurfaces)andusesthePolarisationMeasurementDevices(PMDs)oftheSciamachyinstru-
ment(cp.Sec.1.9.1).ThePMDshaveanoverlappingfieldofviewwiththemeasurementsinthe
mainchannels.Thisisimportant,asthemeasurementsbythePMDsarehenceperformedatthe
sametimeandlocationasthenadirobservationsofchannel3whichareusedfortheIOretrieval.
AsthePMDsmeasureathigherfrequency,thegroundresolutionishigherthanforthespectral
channels.FromthePMDinformation,eachSciamachymeasurementinthesciencechannelsis

79

2Developingtheretrievalofiodinemonoxidefromsatellite

SurfacetypeCloudtypeOther
watericecloudssunglint
cloudsaterwegetationvcloudsgenericw/icesnoland/soildesert

Table2.5:SurfaceandcloudtypesconsideredintheSPICSalgorithm.

classifiedforacertaincategoryandcaninprinciplereceiveuptooneclassificationfromeachofthe
threecolumns(surfacetype,cloudtype,sun-glintyes/no).
Theprincipleofthesurfacetypeclassificationisbasedonthefactthatthereflectivitiesof
differentsurfacetypeshavedifferentspectralcharacteristics.Severalconstraintsandthresholds
areappliedfortheclassifications.RatiosofradiancesmeasuredbydifferentPMDbandsaswell
asthereflectancevaluesatcertainwavelengthsandalsofurtherderivedquantitiesareused.For
detailsthepublicationbyKrijgeretal.(2005)andLotzetal.(2008)shouldbeconsidered.For
eachSciamachymeasurementinacertaintimeperiod,theinformationabouttheallocatedsurface
typeswasprovidedbythedevelopersoftheSPCISalgorithm,W.LotzandM.Vountasfromthe
UniversityofBremen.Withthisdata,testscouldbeperformedforseveralmonthsintheyear2005.
InordertoobservetheeffectofcloudscreeningonthesatelliteIOobservations,allsceneswhich
SPICSclassifiesaswatercloud,icecloudorgenericcloudarediscardedfromtheIOproduct.The
outcomeofthisprocedureshallbecomparedtotheunscreenedresults.InFigure2.15,globalmaps
oftheIOslantcolumnaverageforthethreemonthsofSeptembertoNovember2005areplottedin
thetoppanel.Intheleftmap,thestandardIOproductisshown,andtherightmapcontainsonly
thoseresultswhichhavenotreceivedanypossiblecloudflag.ThescientificanalysisoftheIOresults
themselveswillfollowinChapter3.ThebottompanelofFig.2.15,showsthesameIOproductsas
themapsatthetop,butasacloseupofAntarctica.TheperformanceovertheAntarcticregionis
ofcentralinterestasthediscriminationbetweencloudsandiceisaspecialcapabilityoftheSPICS
algorithm.ThisisimportantforthepresentstudyasonefocusliesonobservationsofIOinthe
Regions.olarPOneimmediateperceptionisthereduceddataamountintherightmaps.Naturally,thesefigures
mustcontainlessdataanditbecomesobviousthatthedatareductionhasanoticeableinfluenceon
thenoiselevelandconsequentlyonthespatialscatter.Themapswithoutcloudscreeningmakea
smoothimpressionincomparisontothemapswherecloudysceneswereeliminated.Apartfromthis
findingthough,themainfeaturespersistalsoaftercloudscreening.IntheSouthernHemisphere
maps,theregionswithhighestIOvaluesagreewellforbothcasesandtheoverallcolumnamounts
lieinthesamerange.Thisobservationalsoholdstruefortheglobalplots.Fromthesimilarityof
thepatternonecanconclude,thatamissingcloudscreeningdoesnotleadtosystematicartefacts
inthestandardproduct.Especially,noappearancesofoutstandingvaluesovercloudyscenesis
ed.observTheexactvaluesofIOcolumnsdifferonlyslightlybetweenthetworesults,whencomparing
theaverageamountsincertainregions.Typicaldifferencesonlyliebelow1·1012molec/cm2.For

80

Maps:Global

Hemisphere:Southern

2.7

screeningCloud

Figure2.15:GlobalmapsandcloseupsoftheSouthernHemispherieforthestandardIOresults
(left)andcloudscreenedresults(right)usingtheSPICSalgorithm.Thespatialnoiseisincreased
noticeablyintherightmaps,apartfromthatthesystematicdifferencesareonlysmall,andno
artefactsareinducedintheleftmapsbynotrejectingcloudyscenes.DuetothenewSPICS
algorithm,itbecomespossiblenowinprincipletousecloudscreenedproductsalsoabovethePolar
Regions.

81

2Developingtheretrievalofiodinemonoxidefromsatellite

tropospherictracegaseslikeIO,especiallywhenthesubstanceisexpectedtobelocatedratherclose
totheground,thevaluesreceivedfromanunscreenedproductshouldbelowerthanaftercloud
screening.Inthepresenceofclouds,apartoftheIOabundanceisshieldedfromthesatelliteview
andtheaveragevaluewouldbereduced.Thereasonwhythedifferencesinthepresentcomparison
arerathersmallmayhavedifferentreasons.However,theconsiderationaboveisgenerallytrueonly
fortracegaseswhichdonotvarymuchintime.IOamountsareusuallytemporallyandspatially
highlyvariablethough,sotheexactvaluesseenbythesatelliteinstrumentcannotbegeneralised
easily.Sceneswhichareclassifiedascloudyoftenstillexhibitpartlycloudfreeregions.Inthe
remainingarea,stillaconsiderableamountofIOmaybedetectedifpresent.Thismeasurement
wouldthenbemissinginthescreenedproductandwouldleadtoacontraryinfluencethanusual.
Inaddition,thepresenceofsomeIOathigheraltitudes,partlyaboveclouds,cannotbeentirely
excluded.Inordertoseedefinitesystematicdifferencesbetweentheuncorrectedandthecloud
screenedproduct,alargerdataamountwouldneedtobeconsidered.Thiswillbedoneinfuture
studies.InSec.2.3thedifferencebetweenbrightanddarksurfacesinthepresenceofcloudshasbeen
mentioned.Piecesofcloudsovericescenesaffectasatellitemeasurementlessthancloudfractions
overdarksurfaceslikeoceans.Fordarkscenes,theintensityfromthecloudypartismuchstronger
leadingtoanoverbalanceofthispartwithrespecttothecloud-freepart,whereasoverice,the
weightingofcloudyandthecloud-freefractionsisaboutthesame.
Theimportantresultfromthisinitialcomparisonisthatnoconsiderableartefactsareintroduced
whenalsothecloudyscenesarekeptintheIOproduct.

2.8Influencingeffectsontheretrieval

IntheprocessoffindingthemostsuitableretrievalsettingsforIOfromtheSciamachydata,
severalparametersneededtobevariedandtheretrievalresultsneededtobeanalysed.Amultitude
ofdifferentretrievalparametersetshavebeeninvestigated,testingtheinfluenceofadditionalor
modifiedfitsettingsandreferencespectra.Fromtheresultsofthesetestrunsespeciallythefit
qualityandtheconsistencyrequirementsasdescribedabovehavebeenevaluated.Someoptionsfor
alternativeretrievalsettingsandtheirrespectiveinfluencesarereportedhere,andoneespecially
importantexampleretrievalisselectedforadetaileddiscussionbelow.

settingsretrievalInvestigated2.8.1Thefollowingchangesandadditionaloptionshavebeentestedintheretrievals.Theutiliseddata
fortheabsorptionfeaturesislistedinTable2.6.

82

•DifferentfittingwindowsincludingdifferentabsorptionlinesofIO
•AdditionaltracegaseslikeO4,H2O(g),and/orGlyoxal(CHOCHO)
•DifferentoradditionalcrosssectiontemperaturesforNO2
•ConsiderationofthevibrationalRamanscatteringinliquidwater

2.8Influencingeffectsontheretrieval

ReferencetsCommenEffectInfluencingAbsorptionbyO4laboratoryspectrumat296KGreenblattetal.(1990)
AbsorptionbyCHOCHOlaboratoryspectrumat296KVolkameretal.(2005)
NO2temperatureeffectlaboratoryspectra,293K&223KBogumiletal.(2003)
AbsorptionbywatervapourlaboratoryspectraHITRANdatabase
Vib.RamanscatteringonH2O(liq)SciatrancalculationVountasetal.(2003)
Table2.6:Additionalatmosphericeffectsontransmittedradiationconsideredonlyinselectedtest
alsretriev

Fromthetestedretrievalsettings,thefittingwindowleadingtolowesterrorsandinterferences
liesaround416-430nm.Inthisregion,theoxygendimerO4andH2Ovapourabsorptionarenot
significantandthereforedon’tneedtobeincluded.Itisinevitabletoincludethesegaseswhen
usingwavelengthwindowsexceeding440nm.TheconsiderationofCHOCHOintheretrievalleads
toapparentlyincorrectresultsforhighlatitudes,alsothestrongestabsorptionlinesliebeyondthe
selectedwavelengthwindow.
ThedifferenttemperaturesofNO2showlittleimpactonthefitqualityandtheIOamounts
inthechosenspectralrange,thereforeonlyonetemperatureistakenintoaccountinthestandard
fit.Insomeretrievals,theVRSshowedthewrongalgebraicsigninsomelocations.Inthecurrent
fittingwindow,VRSdoesnotshowalargeinfluenceontheretrievalandisthereforeomittedinthe
duct.prostandardWhenselectingtheretrievalparameters,thenumberofvariablesinthefitshouldbekeptsmall
inordertoavoidover-determinationofthesystemofequations.Especiallyforsmallwavelength
windowsthisisanimportantissue.Fromthenumberofspectralpointsinthemeasurements
(correspondingtodetectorpixels)andthewidthoftheslitfunction,thenumberofindependent
piecesofinformationisdetermined.Thenumberoffreevariablesinthefitshouldnotexceedthis
valueandalsoshouldnotlietoocloseduetotheremainingnoiseinthemeasurements.Therefore,
thenumbersoffittingparametersiskeptassmallaspossible.
Ofalltestedparametercombinations,someretrievalresultsandretrievalfailuresofoneselected
parametersetarepresentedinthenextparagraph.Thisspecificretrievalischosenasexamplehere
forseveralreasons.Straightforwardandlikelychoicesofparametershavebeenmade,soinitial
retrievalattemptsusedparametersetssimilartothisone.Theretrievalresultsshowinteresting
errorfeaturesandcorrelations.Furthermore,theconsiderationsconnectedtothisretrievalareof
importanceforlatercomparisonstoanindependentstudy(cp.Sec.4.2).

2.8.2Retrievalinthe418-438nmwindow
TheretrievalversiondiscussedinthisparagraphisVersion0.27i.Inprincipleitwouldbedesirable
toselectafittingwindowthatincludesatleastthethreemostintensiveabsorptionfeaturesfromthe
absorptioncrosssectionspectrumofIO.Thesearetheabsorptionbandsfrom(ν=3←ν=0)
to(ν=5←ν=0)oftheA32/2←X32/2transition,cp.Fig.1.4,andliebetween418and438nm.

83

2Developingtheretrievalofiodinemonoxidefromsatellite

Inseveraltestrunstherefore,thisfittingwindowwaschosenfortheretrievalofIO.

SettingsarameterPFittingwindow418-438nm
AdditionaltracegasesNO2temperatureeffect(293K-223K)
absorptionO4vibrationalRamanscatteringinH2Oliq
Table2.7:FitsettingsfortheproblematicretrievalV0.27idifferentfromthestandardfit.

Theresultsforthiswindowshowedsomeproblems,whichdonotallowproperIOretrievalinthis
spectralrange.Fortheproblematicretrieval,therelevantsettingswhicharedifferentfromthe
standardfitarelistedinTab.2.7.FirstevidenceofirregularitiesinthefitcanbeseeninFig.2.16.
GlobalmapsoftheretrievalresultsforIOfromtheproblematicretrievalsettingsareshown,aver-
agedoverthemonthofOctober2005.Intheleftmap,thefitqualityislimitedcorrespondingto
anopticaldepthrmsof5×10−4,andsomeregionswithhighIOslantcolumnsarevisible.After
applyingamorestrictcriterionofrms<2×10−4inFig.2.16(b),theresultslookcompletely
different,asallhighIOresultsvanish.Acorrelationbetweenlowerfitqualityandthedetection
ofhighIOamountsisthusidentified.Thisisthemostdirectevidenceforfundamentalretrieval
difficulties.Anadditionalobservationisthattheregimeofspecificfitparametershereishighlyunstable
withrespecttocertainfitsettings.Insimilarretrievalruns,whereonlyminorsettingchangeswere
applied(e.g.adifferentchoiceofthecalculatedVRSspectrum),theIOsignaturesfliptheirsign,
i.e.,locationswithhighIOinFig.2.16(a)thenexhibitstronglynegativeIOslantcolumns.The
spatialcorrelationofIOandhighrms-valueintheglobalpictureremainsinallsimilarretrieval
runs.

(a)

(b)

Figure2.16:GlobalmapsofIOresultsfromtheretrievalintheproblematic418-438nmwindow.
Thetwocasesshowresultsafterapplicationofdifferentlystrongqualitycriteria.Acorrelationof
lowerfitqualityandhigherIOamountsisidentified.

84

(a)

(b)

2.8Influencingeffectsontheretrieval

Figure2.17:(a)showsthespectralIOfitresultwiththescaledlaboratorycrosssectioninblue
andthefitresultincludingtheresidualspectruminblack.ThedetectionofahighIOamountis
theconsequenceofastronglystructuredresidual.(b)showstheresultfromthesamemeasurement
forthepoorfittingoftheRingeffect.Itbecomesclearthatseveralfeaturesinthespectrumare
captured.erlypropnot

resultsfitectralSpAspectralfitresultofatypicalfailedretrievalisdisplayedinFig.2.17forameasurementwhere
highIOwasdetected.Thefigurecontainstwocurves,thebluecurveisthescaledlaboratory
spectrumoftheabsorptioncrosssection,theblacklineistherespectivemeasurementfitresultof
theabsorptionspectrumcontainingtheresidualspectrum.Strongspectralstructuresappearinthe
residual,whichareclearlynotrelatedtopurenoise.Partly,theIOcrosssectioncoincidesspectrally
withotherwiseunassignedstructuressothatanIOsignalispickedup.Fromthispoorfitresult,it
mustbeconcludedthattheretrievedIOamounthereistheresultofanimproperretrievalprocess.
TheshapeofthefitresidualdisplayedinFig.2.17(a),occurspersistentlywheneverespecially
highIOamountsaredetected.Suchastable,highlystructuredresidualindicatesanunderlying
fundamentalproblem.ThemostprobablecauseherewasidentifiedtobethefittingoftheRing
effect.TherespectivefitresultfortheRingeffectisdisplayedinFig.2.17.Thedifferencesbetween
thefitandthecalculatedspectrumareespeciallystrongatthepronouncedfeaturesoftheRing
effect.Inmanyplaces,eitherthepeaksortheminimaarenotwellcaptured.

Circumstancesforsceneswithlargefittingerrors
ThefactthathighIOamountsareconnectedtothepersistentstructure,isdemonstratedin
Figs.2.18-2.20.ThevisualcolourfromthePMDmeasurementsisshowninFig.2.18(a),where
cloudsappearinwhite,forthe1stofSeptember,2005,overEastAsiaandtheIOslantcolumn
resultsontheright(b)forthesametimeandscene.Inseveralcases,thehighIOamountscoincide
withtheoccurrenceofclouds.ThecentralmeasurementstateoverAsiafromFig.2.18ispickedout
inFig.2.19withfoursinglemeasurementsselectedandnumberedforcloserinvestigation.Fig.2.20
showsthefitresultsoffourSciamachymeasurementsincomparison.Themeasurementsnumber
1and3arenotstronglyaffectedbycloudsandtheresidualoftheretrievaliscomparablysmalland
onlyslightlystructured.Formeasurementsnumber2and4,wherebrightcloudscoverthefieldof

85

2Developingtheretrievalofiodinemonoxidefromsatellite

view,theresidualissignificantlylargerandstronglystructured.Inparticular,thestructuresare
strikinglysimilarinbothcases.Obviously,asystematicirregularityintheretrievaloccursabove
brightsurfacessuchascloudsandice(Fig.2.16),prohibitingreasonableIOretrievalwiththese
settings.parameter(b)(a)

(b)

Figure2.18:SatellitemeasurementsforwhichspectralfittingresultsareshowninFig.2.20.(a)
showsthePMDvisualimpression,wherecloudsareseeninwhite,and(b)givestheIOresultsfrom
theproblematicfit,whereinmostcases,highIOvaluesarecorrelatedwithcloudyscenes.

Possiblephysicalreasonsforthefiterrors
ThesystematicoccurrenceofthestableresidualshowninFigs.2.17and2.20asksforexplanations
ofitsorigin.TheIOresultsinFig.2.16and2.18showtherelationofIOdetectioninsceneswith
eitherbrightcloudsorbright(ice)surfaces.However,notallicecoveredregionsandcloudypixels
areaffected.Additionalcircumstancesmustplayarole.
Thespectralstructuresinthestrongresidualsareprominentaroundthe430-431nmregionand
vanishifthefittingwindowisrestrictedtowavelengthsbelow.Thishintsatapossibleconnection
betweentheinaccuratefittingandtheFraunhoferGband(cp.Fig.1.7).Beingthemostprominent
Fraunhoferfeatureintheconsideredfittingwindow,theGbandalsoexhibitsthestrongestinfilling
(Sec.1.6.3).TheRingeffectmustbewelldescribedatthisspectralposition,otherwiseerrorsare
induced.OnepossibleexplanationisthattheRingeffectspectrumusedintheretrievaldoesnot
capturethispositionwellincertainsituations.InadditiontotheRingeffect,theFraunhoferlines
playarolealsoforothereffects,liketheimpactofinstrumentalstraylightonthemeasurementsor
thepolarisationsensitivityoftheinstrument.

86

2143

Figure2.19:Sciamachystatefromthe
mapshowninFig.2.18withfournum-
beredgroundpixels,forwhichretrieval
resultsareshowninFig.2.20todemon-
stratetheconnectionofincorrectfitting
withcloudcover.

2

4

2.8Influencingeffectsontheretrieval

1

3

cloudFigurefree2.20:scenes,FittheresultsIOforvatheluesfouraresmallgroundandscenesthenumbresidualeredinlargelyFig.2.19.unstructured.Cases1andCases3are2abandove4
sphowevectrum.er,areThesechosenovsystematicercloudyeffectssatelliteleadtopixelsimpropanderbothfittingshowandtheerroneoussamehighlyresultshere.structuredresidual

TheRingeffectchangesitsshapeandstrengthasconsequenceofseveraleffects.Onestraightfor-
wardinfluenceofcloudsistheshieldingofthelowestpartoftheatmosphere,wheremostscattering
moleculesaresituated,therebyreducingtheRamanscatteredportionintheradiation.Theinfilling
ofabsorptionlineswillbesmallerincaseofahighcloud.Thiseffectcanactuallybeusedtoderive
thecloudtopheightfromsatellitemeasurements(JoinerandBhartia,1995;deBeeketal.,2001).
AlsoabovebrightscenessuchasiceontheEarthsurface(whichisasimilarsituationtoverylow
clouds)theRingeffectisslightlyweakerthanoververydarksurfaces,astheRamanscatteringis
overbalancedbyLambertianreflectionandinfillingissomewhatreduced.Inthecaseofabright
cloud,ifpartoftheatmosphereisshieldedandallotherparametersarekeptconstant,themajor
influenceontheRingeffectisachangeofthestrength.Inaddition,thespectralshapeofthe
Ringeffectmaychangeindependenceofthesurfacealbedo,asthealbedoimpactsonthetypical
lightpathlengththroughthesurfacelayers.Differentwavelengthswillthenexperiencedifferently
strongRamanscattering.Thiseffectisnottakenintoaccountintheretrieval,asitdoesnotlead
toanynoticeableproblems.TheRingeffectusedintheretrievaliscalculatedforanalbedoof90%,
andtherearenosystematicirregularitiesfordarksurfaces.Also,theproblemsoverbrightscenes

87

2Developingtheretrievalofiodinemonoxidefromsatellite

arerestrictedtocertainsituations.TestswithRingspectracalculatedforvaryingconditions(e.g.,
changingsurfaceheight,SZA,groundspectralreflectance)didnotimprovethesituationinthelarge
wavelengthregion.ApureinfluencefromtheRingeffectassinglecauseforthefittingdifficulties
inthe418-438nmwindowcalculationisregardedunlikely,mainlyduetotheselectiveappearance
oftheproblem.However,acombinedinfluenceoftheRingandothereffectsmightbecomerelevant.

Oneaspectwhichisaffectingbothsituationssimilarly,thecloudcaseandthebrightsurface,is
theenhancedstraylightinfluence.Thisoccursespeciallyifaspectralpartoutsidethefittingwindow
isincreased.Abovevisuallybrightsurfaces,thelongerwavelengthpartofthespectrumisenhanced
relativetotheshorterwavelengthsascomparedtodarkersceneswheretheshorterwavelengths
dominateduetotheλ−4-dependenceoftheRayleighscatteringprobability.Thespectralshapeof
thestraylightinfluenceresemblesthatoftheRingeffect(andisthereforemostprominentatthe
Fraunhoferlinepositions),whichmaybeillustratedinthefollowingway.TheintensityIrecorded
bytheinstrumentisgeneratedbythesumofelasticandinelasticscattering:I=Iel+Iinel,and
ln(I)=ln(Iel)+ln(1+IIelinel).ForthecaseofdominantelasticscatteringI≈Iel>>Iinel,the
followingapproximationisvalid:
ln(I)≈ln(Iel)+IIinel≈ln(Iel)+f·I1
elInthelaststep,byapplyingaconstantfactorf,thewavelengthdependenceofIinelisneglected
incomparisontothewavelengthdependenceofIel,whichisstrictlynottruebutmaybeusedas
anapproximation.ConsideringtheDOASfittingmethod,thiscorrespondstousingI1aspseudo
absorberfortheRingeffectandfastheassociatedfitfactor.BeforetheRingeffectwascalculated
byradiativetransfermodels,theinverseoftheradiancewasusedtoaccountfortheinfillingin
ground-basedmeasurements(JohnstonandMcKenzie,1989;Noxonetal.,1979).Althoughthisis
onlyaroughapproximation,thesimilaritytothestraylightapproximationinSec.1.8isappar-
ent,forthecaseofconstantstraylightamount.Duetotheserelations,spectralfeaturesatthe
positionsoftheFraunhoferlinesarealwaysmostprominentinbotheffects.Theresultingspectral
correlationbetweenRingandstraylighteffectmaygeneratefittingerrorsifoneorbotheffects
arestrongandnotpreciselyaccountedfor.Possibly,theconsiderationofthestraylightisnot
sufficientinthepresentlyimplementedway.However,thedistinctivefeaturesintheretrievaldo
notoccuraboveallbrightsceneswherestraylightislarge,butonlyoversomeofthem.Ifstray
lightisonecauseoftheerrorfeatures,againthereneedstobeacombinationwithafurtherinfluence.

AthirdprocesswhichleadstospectralstructuresattheFraunhoferlinesisthepolarisation
sensitivityoftheinstrument.Ingeneral,lighttransmissionthroughagratingspectrometeris
polarisationsensitiveduetopolarisationdependenciesoftheopticalcomponents(mirrors,grating).
Therefore,therecordedsignalatthedetectorwilldependonthepolarisationdegreeoftheincident
radiation.ThepolarisationstateoftheradiationfieldisdescribedbytheStokesvectorS=
(I,Q,U),whereIisthetotalintensity,Q=Ip−Isistheintensitydifferencebetweenhorizontal
andverticalpolarisation(withrespecttothespectrometerentranceslit)andU=I+−I−is

88

2.8Influencingeffectsontheretrieval

theintensitydifferenceoftheradiationpolarisedatanglesof±45◦(Gottwaldetal.,2006).The
polarisationsensitivityofasystemmaybespecifiedbyaMuellermatrixM(Hecht,2001;Gottwald
etal.,2006).Withtheradiationtermsatthedetectormarkedwithsubscripteddetandtheincident
radiationtermswithin,thetransmissionofradiationthroughtheinstrumentfollowstheequation:
⎛I⎞⎛I⎞
⎝⎜Q⎠⎟=M·⎝⎜Q⎠⎟
UUindetThefirstentryinthedetectorStokesvectoristheactuallyrecordedintensityIdet,whichentersthe
DOASanalysisprocedures.WiththematrixelementsMijtherecordedintensityreads:
M12M13QU
Idet=M11Iin·1+M11q+M11uwithq=Iin,u=Iin
Inthefollowing,theconsiderationsarerestrictedtothedependenceonqforsimplicity,asthe
u-dependencyisequivalent.Thequantityqisthestandardpolarisationdegree.Foropticalsensors
oftenthefunctionηisdeterminedinpre-flighttests,fromwhichthematrixelementsM11andM12
arecalculatedbyMM1112=11+−ηη.Consideringtheopticaldepthbetweentherecordedintensityandan
referenceintensityI0,whichisthetermanalysedintheDOAS-algorithm,thiscanbewrittenas:
IdetIin1−η
lnI0=ln(M11)+lnI0+ln1+1+ηq(2.1)
Iin1−η
≈ln(M11)+lnI0+1+ηq(2.2)
Theapproximationholdsifpolarisationisnottoostrong,otherwisetheimpactisevenmorepromi-
nent.ThelasttermintheequationdependsontheinverseintensityIin−1andisthereforesimilarto
theeffectsdiscussedabove.However,thewavelengthdependenceofthefactorinfrontoftheinverse
intensityandespeciallythatofQinarecomparablystrongandoftencannotbeapproximatedbya
constantorlinearfunctionasthisisdoneforthestraylighteffect(cp.Sec.1.8).Furthermore,Qin
istypicallynotknown.
Thepolarisationdegreestronglydependsontheratioofthedifferentscatteringtypes(Rayleigh,Ra-
manandMie),becauseRayleighscatteringisstronglypolarising,Ramanscatteringisonlyweakly
polarisingandMiescatteringisnotpolarising.Thepolarisationdegreeofthebackscatteredradi-
ationthereforedependsontherespectivescene.Overclouds,e.g.,Miescatteringisstrong.The
referencespectrumI0overthePacificistakenoveraclearscene,sothatmainlyRayleighscat-
teringcontributes.Thepolarisationdifferencebetweenthereferencesceneandthebrightscenes
overcloudsandicemayleadtoadditionalstructuresinthespectrumwhicharenotdescribed
bytheappliedcrosssections.Inaddition,thedegreeofpolarisationalsodependsonthescatter-
ingangle,i.e.ontheviewingangleofthesatelliteandthepositionofthesun.Thismayexplain
theobservation,thatthelargeresidualsovercloudsandbrightsurfacesonlyoccurinsomelocations.

Inconclusion,thefittingdifficultiesinwindowscoveringtheFraunhofer-Glinearound431nm

89

2Developingtheretrievalofiodinemonoxidefromsatellite

arepossiblygeneratedbyacombinationofseveralinfluenceswhicharestrongatthiswavelength
position.SpectralcorrelationsoftheIOspectrumwithremnantsoftheG-band,theappliedcorrec-
tionspectraforRingandVRSthencauseproductswhicharemoretheresultofmutualcorrelations
thanofrealatmosphericconditions.Themostprobableimpactresultsfromthepolarisationsen-
sitivityoftheinstrumentandmaybeimprovedeitherbyabetterpolarisationcorrectionofthe
Sciamachyrawspectraorinfutureinstrumentsbyemployingapolarisationscrambleratthe
sensor.theoftranceen

Consequenceforthechoiceoffittingwindow
Inthe418-438nmrange,nofitparametersethasbeenfoundsofar,wherethedeficienciesatthe
Fraunhofer-Glineareproperlyeliminated.Thefiteithertendstostronglynegative(notshown
here)orstronglypositive(cp.Fig.2.16)IOamounts.Theseproblemsneedtobesolvedbeforethe
retrievalofIOinaspectralwindowspanningmoreIOabsorptionlinescanbecomesuccessful.Asa
consequenceofthedifficultiesinthe418-438nmwindow,theIOretrievalhasbeenrestrictedtothe
spectralrangeof416-430nm,yieldingthestandardIOretrieval.Thisway,thediscussedproblemof
strongresidualsresultingfromapoorfitaround431nmisavoided.AnevenbetterretrievalofIO
mightbeachieved,ifthemeasuredradiancesinthisspectralrangewerebetterunderstood.Future
investigationsshouldconsiderthepolarisationsensitivityoftheinstrumentasonelikelycausefor
thestronglystructuredandlargeresidualsincertaincloudoricescenes.Assoonasthequestionis
solved,whichinfluencesexactlyprovoketheretrievalinaccuracies,newretrievalsinalargerfitting
windowmaybeconducted.

2.9Thechoiceofthebackgroundspectrum
ConsiderationsinthissectionuseresultsfromtheIOfitVersionV2.54(cp.Sec.2.2.1).Forthis
dataproduct,thelargePacificregionshowninFig.2.21ischosenasreference,overlappingthe
PacificreferenceregionofthestandardV1.28product(cp.Fig.2.4),butbeinglargercausinga
higherdegreeofaveraging.

90

cificFigurereference2.21:Mapregionsho(Vwingersionthe2)largeusedPain-
thisgionforsectiontheIOandasdatavestandardrsionV2.54.referencere-

FortheperformanceofaDOASretrieval,thechoiceofthebackgroundspectrum(I0)maybe

2.9Thechoiceofthebackgroundspectrum

crucial.Whileasolarreferencespectrumcontainingnoatmosphericabsorbersignatureswouldbe
oneappropriatechoice,thismaycausedecreasedretrievalqualityinsomecases.Thisisdueto
tworeasons-thenoiselevelinSciamachysolarspectraishigherthaninanaveragedearthshine
spectrumandadditionallythepossibilityforretrievalartefactsishigher.Thiscanbecaused,e.g.
byalargespectralinfluencefromtheRingeffect,whichisnotpresentindirectsunlightbutis
prominentinEarthshinemeasurementsofscatteredandreflectedsunlightinspectralwindowswith
strongFraunhoferlines(cp.Sec.1.6.3).Inaddition,theimpactofsomeremainingcalibration
inaccuracies,whicharesimilarforindividualEarthshinescenesbutmaybequitedifferentforthe
Solarspectra,maybereducedthisway.WhencomparingtwoEarthshinespectra,andformingtheir
ratiointheDOASretrieval(cp.Sec.1.8),partoftheseeffectscancelsout.Therefore,thechoiceof
anappropriateEarthshinespectrumasbackgroundusuallyleadstothemoststablefitresultshere.
Asinstrumentalcharacteristicsandparametersmightexperiencetemporaldrifts,abackground
spectrumcloseintimeoftenisanecessarychoice.Additionally,thelocationofthisbackground
spectrumhastobechosencarefully.Thebestchoiceisalocation,wherethespecifictracegasunder
investigationisnotpresentor,alternatively,whereitsamountiswellknown.Insteadofusinganew
specificallygeneratedEarthshinereferencespectrumeveryday,amodifiedprocedurecanbeused.
AconstantEarthshinespectrumfromafixedtimeandplaceisappliedandinasecondstep,the
influencesofinstrumentaldriftsontheretrievalareeliminatedbysubtractingtheaverageddaily
fitresultinthebackgroundlocationfromtheindividualmeasurement.Themathematicalbasisfor
thisprocedureisgivenbythefollowingequation:
I(ti)I(ti)I0(ti)
lnI0(ti)=lnI0(tc)−lnI0(tc),(2.3)
whereIandI0aretheintensitiesmeasuredatanylocationontheEarthandinthebackground
region,whiletiandtcdenoteanytime(day)ofmeasurementandthefixeddayoftheconstant
backgroundspectrum,respectively.Thesecondtermontherighthandsideln(II00((ttci)))describesthe
differencesinthebackgroundspectrumatdifferenttimesandmaycontainabackgroundoffsetof
thetracegasofinterest.Thelefthandsideoftheequationabovereferstotheretrievalwithdaily
reference,whiletherighthandsideusesthefixedreferenceincombinationwithsubtractingthefit
resultinthebackgroundregioneveryday.Thesecondapproachisusedinmostofthefitsincluding
thestandardfitandfromthebasicprinciple,thisprocedureshouldbefunctional.Forthereal
retrievalthough,itshallbeshownexemplarilythatthereisnobigdifferencefortheIOretrieval
resultbetweenthetwomethods.
ThepureVersionV2.54C(tagC)usesaconstantbackgroundEarthshinespectrumfromthe
1stSeptember2007,extractedfromwithinthereferenceregion.Foreverydaynormalisationtothe
referenceregion,thedailyaverageofIOinthereferenceregionissubtracted.Thefinalproductis
thenreferredtoasVersionV2.54C-AVE.Theequivalentfitparametersetwiththeonlydifference
ofusingadaily(tagD)referenceisVersionV2.54D.Thesetworetrievalsettingsshallbecompared
wing.follotheinThecomparisonisnowperformedfortwodifferentdays,the20thSeptember2006(different
yearthanbutsameseasonastheconstantbackgroundspectrum)andthe9thMarch2006(different
yearanddifferentseason).Forthesedays,allrecordedSciamachyspectrawithinthereference

91

2Developingtheretrievalofiodinemonoxidefromsatellite

regionareaveragedtoyieldreferencespectraforthespecificdaystobeusedinthefitV2.54D.
ComparisonsoftheresultsareshowninFig.2.22,wheretheIOslantcolumnresultsfromtheentire
satelliteorbit(20060920_183and20060309_190)areplottedversustherespectivelatitudeofthe
measurement.TheIOfitresultswithdailyreference,IOD(lat),aremarkedbyredsquares,theblue
trianglesshowtheretrievalusingconstantreference,IOC(lat),andsubsequentoffsetcorrectionby
theaverageIOinthereferenceregion,IOC,AVE.Differencesbetweenthetwomethodsaresmall
andcannotbedistinguishedinthisgraph.
(a)2.22:FigureResultingIOslantcolumnamountsfor
twosampleorbitsinMarch(a)and
September(b),2006.Ineachgraph,
theresultsoftwodifferentmethodsare
compared,whichdealwiththeEarth-
shinebackgroundspectrumindifferent
wausesys.theRetrievdailyaalverageV2.54D(redEarthshinesquares)from
thebackgroundregion,whilethever-
sionshowninbluetrianglesstusesacon-
stantreferencefromthe1Septem-
(b)ber,2007.Forthelattercase,thedaily
averageIOamountIOC,AVEintheref-
onderencestep,regionyieldingisthesubtractedplottedinvaersionsec-
V2.54C-AVE.Thetwomethodsyield
resultswhichareequalwithinsmallde-
viations.

(b)

DayGlobalmeanΔ¯Standarddeviationσ(Δ)Referenceoffset(IOC,AVE)
-5.740.06-0.0920.09.2006-7.110.100.8609.03.2006TandableVersion2.8:V2.54DStatistical(Equationinformation2.4).onThetheunitdifferenceofthevΔaluesinIOis10results12molecbetw/cmeen2.VersionV2.54C-AVE

ThedifferenceinretrievedIObetweenthetwomethodsiscalculatedbyΔ(lat),wherelatis
latitude:theΔ(lat)=(IOC(lat)−IOC,AVE)−IOD(lat)(2.4)
SomestatisticalinformationforΔ(lat)fromtheinvestigatedorbitsisgiveninTab.2.8.Remaining
deviationsontheorderof<1×1012molec/cm2(often<1×1011molec/cm2)aremainlycausedby

92

2.9Thechoiceofthebackgroundspectrum

statisticalfluctuationsandretrievaluncertainties(cp.Sec2.5).Overall,thedifferencebetweenthe
twomethodsisonlysmall.Neitherthetracegasresultsnorthefitqualitiesshowspecificdeviations
problems.orOneadvantagemaybethereducedriskoferrorwhenusingaconstantreference.Betterinspection
ofthefitresultispossible.Potentialirregularitiesinthereferenceregioncanbeseeninthepure
resultbeforesubtractionoftheaveragedreferenceoffsetandwillnotaffecttheglobalresults.
However,bothmethodsareregardedassuitableaslongasnosuddeninstrumentalchangeinthe
timeseriesoccurs.Inthatcase,adaily(orrenewedconstant)Earthshinespectrumisnecessary.
Currently,themethodofusingaconstantEarthshinereferenceisretained.

93

3ObservationsofIOfromsatellite

AftertheretrievalofIOfromSciamachydatahasbeenoptimised,thedetectionofenhanced
amountsofIOinsomeregionsonEarthbecomespossible.DuetothefactthatabundancesofIO
arecomparablysmallandoftenlieclosetothedetectionlimit(Sec.2.4)ofthesatelliteinstrument,it
isnotpossibletoobserveshorttermtransporteventsorthelocalisationofemissionsoncertaindays.
Itisnecessarytoaveragetheresultingslantcolumnvaluesovercertaintimeperiods.Temporal
averagingalwaysmeansatradeoffbetweenanimprovementofsignal-to-noiseandalossoftemporal
resolution.Withanaveragingperiodofthreemonths,e.g.,thenoiseisconsiderablyreducedas
comparedtomonthlyorevendailyresults,butsometemporalvariationcanstillberesolved.The
strategyandperiodofaveragingwillbeineachcaseadaptedtothespecificsituationorobjective
analysis.ofInthefollowing,globalresultsforIOobservationsusingthecurrentlymostconsistentretrieval
settingswillbepresented.AllpresentedmapsandtimeseriesofIOcolumnsinthischaptershow
datafromVersionV1.28discussedinSec.2.2,whichhavepartlybeenpublishedinSchönhardtetal.
(2008).Alongsideanoverviewovertheglobalresults,severalinterestingregionsarehighlighted.
AsonemainfocusofthisstudyliesoniodineinthePolarRegions,IOamountsandseasonal
variationsoverAntarcticaarediscussed.Thenewfindingsarepresentedandpossibleapproaches
forinterpretationareexplained.AfurtherinterestingregiondiscussedinthischapteristheEastern
PacificandfinallysomelocationsontheNorthernHemisphere.

observationsGlobal3.1

Asanoverview,completeglobalmapsforcertaintimeperiodsareconsidered.Theglobalmapsallow
aviewalsoontoregionswherenoiodinemeasurementshavebeenundertakenbefore.EnhancedIO
amountsabovetherespectivedetectionlimitarefoundincasesofwidespreadsourcesandfavourable
conditions.Figure3.1showstheaverageofallretrievedSciamachyIOobservationsoveratimeperiodof
fouryearswithaspatialresolutionof0.25o×0.25o.Theslantcolumnamountisshownincolour
code,whichisdisplayedontheright.Thecolourcodeappliedhereisthesameasintroducedin
Sec.2.6andwillbeconsistentlyusedthroughoutthisstudyforthestandardIOresults,sothatall
plotscanbedirectlycomparedwitheachother.
OverlargeareasoftheEarth,IOcolumnsarebelowthedetectionlimit.Followingthedetec-
tionlimit,precisionandaccuracyconsiderations(Sec.2.4and2.5),mainlyIOcolumnsexceeding
3·1012molec/cm2areinvestigatedandfurtherdiscussed.Regionswithsufficientlylonglasting
andwidespreadenhancedIOvaluescanbedistinguished.Inthisaverageoverseveralyears,the
maximumvaluesreachuptoabout7·1012molec/cm2.Thelargestandwidestspreadvaluesare

95

3ObservationsofIOfromsatellite

foundclosetoandoverAntarctica,especiallyintheWeddellSea,overshelficeregionsandat
coastlines.Therehavebeenground-basedmeasurementsinAntarcticaalsoshowingenhancedIO
amountsincertaintimesoftheyear,whichwillbediscussedinmoredetailinSec.4.1.

Figure3.1:GlobalslantcolumnamountsofIOaveragedfromJune2004untilMay2008.

SeasonalplotsaveragedoverfouryearsbasedonthesamedataaredisplayedinFigs.3.2(a)-(d).
Itbecomesvisible,thattheIOvaluesaroundAntarcticaarenotconstantovertime,butexhibit
annualvariations.ThesefindingsaresubjectofSection3.2and3.3.
ApartfromtheprominentSouthpolarregion,alsootherareasexhibitenhancedslantcolumnvalues
ofIO.LookingattheequatorialEasternPacificNorthwestofSouthAmerica,enhancedIOvalues
areobserved.MoredetailsonthisregionfollowinSec.3.4.
Notvisibleinthemapspresentedhere,therearealsonotionsofenhancedIOvaluesontheNorthern
Hemisphere.Specialaspectsintheretrievalplayarolefortheresultsthereandneedtobediscussed
3.5).Sec.(cp.carefullyTwoirregularitiesneedsomeattention.First,theIOvaluesshowatendencytosmallerand
evennegativevaluestowardsandontheNorthernHemisphere.Thisfacthasinfluencesonthe
choiceofthereferenceregionwhenanalysingtheresultsinNorthernregionsandwillbediscussed
3.5.Sec.inAsecondpointofconcernaretheclearoceanregionsmainlyinthePacificandalsointheAtlantic
andIndianOcean.Inareaswithoutbiologicalproduction(oceandeserts)wherethewaterisvery
clear,influencesontheretrievalsofseveraltracegasesoccur,probablybylightthattravelsthrough
thewaterforacertaindistancebeforebeingbackscatteredintospace.ForIO,whichisaminor

96

(a)

(c)

(b)

(d)

3.2ObservationsofIOinAntarctica

Figure3.2:GlobalslantcolumnamountsofIOaveragedfordifferentseasons.Takenfromthe
overalltimeofJune2004toMay2008,theperiodscover(a)March-Mayfor2005-2008,(b)June-
August2004-2007,(c)September-November2004-2007,(d)December-February2004-2008.

absorberandwhereamountsareusuallyveryclosetothedetectionlimit,theinterferenceoflight
withprocessesinwatermayhaveastronginfluenceandcausesystematicnegativevaluesoverthese
oceandeserts.ThisperceptionhasledtotheanalysispresentedinChapter6.

3.2ObservationsofIOinAntarctica

SomelocalmeasurementsinAntarcticahavereportedobservationsofIObefore(Friessetal.,2001;
Saiz-Lopezetal.,2007b)andalsooforganiciodinecomponents,e.g.CH2I2(Reifenhäuserand
Heuman,1992;Carpenteretal.,2007).MeasurementsofiodinespeciesintheAntarcticaregenerally
sparseanditisamajorinteresttoinvestigatethisareafurther.Observationsfromspacemaybe
ofgreatsupportinordertoanalyseamountsandvariationsofIOandalsopossibleconnections
betweenIOandBrO,aswellasbetweenIOandicecover.

97

3ObservationsofIOfromsatellite

3.2.1SeasonalvariationofIOinAntarctica
FortheobservationsofIOinandaroundAntarctica,itisconvenienttolookatpolarmaps.The
graphicsinFig.3.3covertheSouthernHemisphereupto40◦Southernlatitudeandshowseasonal
averagesoftheIOslantcolumnsfortheperiodsMarch-May,June-August,September-November
andDecember-February.TheoveralltimeperiodofthedatashownisJune2004toMay2008,
andeachmaprepresentsanaverageover4subsequentyearssothatthenoiselevelisstrongly
reduced.ThetemporalvariationsinthedistributionofIOvaluesinandaroundAntarcticashow
similarsignatureseveryyearwhichwillbeseenlater.Therefore,theresultscanbeaveragedover
subsequentyearsandmeaningfulconclusionscanstillbedrawn.
StartinginAustralspring(i.e.fromaroundSeptemberintheAntarctic),largevaluesofIO
occurintheWeddellSea,inshelficeregionsandalongtheAntarcticcoastlines(Fig.3.3a).Also
ontheAntarcticcontinent,positiveIOvaluesareapparent.Thelargestaveragevaluesreachup
toabout8×1012molec/cm2.Singlemeasurementscanbeconsiderablyhigher,e.g.,afitresultof
2×1013IOmolec/cm2waspresentedinChapter2.TheIOamountsarethenlowerinthesummer,
butstillremainabovedetectionlimitinsomeareas.Anincreaseoftheamountsandthecovered
areatowardstheautumnperiodagaincanberecognised.InSouthpolarwinter,nolightisavailable
forthistypeofsatelliteobservations.Regionswithoutdataappearwhiteintheimages.Butalsoat
therimofthemeasurementregion,nosystematicallyenhancedvaluesareobserved.Atthisborder
ofthemeasurementregioninwintertimeandalsoinsomeextentfortheautumntime,alarger
amountofscatteringisapparentinthedata.Thisisduetoasmalleramountofdatapointswhich
fulfillthecriterionofhavingaSZAbelow84◦.
SeveralaspectsofthisgeneralseasonalvariationfoundinthesatelliteIOobservationsshallbe
analysedinmoredetailnow.

3.2.2IOtimeseriesatHalleyStation,Antarctica
Theannualcycleroughlydescribedabovemaybeobservedinmoredetailwhenlookingattime
seriesforcertainlocations.AtimeseriesextractedfromaroundHalleyStation(cp.Fig.3.4),
situatedinAntarcticaat75.5◦S,26.5◦WisshowninFig.3.5.Datafromthistimeseriesformsthe
basisofaveryimportantvalidationstudypresentedinSec.4.1wheresatelliteandground-based
dataarecompared.ForthedatapointsinFig.3.5,allsatellitemeasurementswithinaboxof
500kmsidelengthenclosingHalleyStationhavebeentakenintoaccountandareaveragedoverone
dayeach.ThesedailyIOslantcolumnsareplottedinblueandaweeklyrunningmeanisoverlaid
k.blacinTheevolutionoftheIOvaluesisshownforfoursubsequentyearsandineachyear,asimilar
developmentisobserved.LargestIOamountsalwaysappearshortlyafterpolarsunriseinSeptember
withhighestvaluesinOctoberaround6to7×1012molec/cm2intherunningmean.Aftera
slightsummerdecreasedowntovaluesaround2×1012molec/cm2,theautumnagainexhibitslarger
amounts.Towardswinter,theIOindeeddecreasesagain,andduringwintertimenodataisavailable
darkness.todueInpart,IOslantcolumnsinsummerarelowerthaninspringandautumnbecauseofahigher

98

(a)

(c)

(b)

(d)

3.2

ObservationsofIOinAntarctica

Figure3.3:SeasonallyaveragedslantcolumnamountsofIOabovetheSouthernHemisphere(up
to40◦S)fromAustralspringtowinter.ThedataistakenfromJune2004toMay2008forthe
periodsofSeptember-November(2004-2007),December-February(2004-2008),March-May(2005-
2008),andJune-August(2004-2007).MaximainIOcolumnsoccurovertheWeddellSea,theRoss
Seaandalongthecoastespeciallyinspringandinautumnwithlowerlevelsyetremainingpositive
atsomeareasthroughoutthesummer.

99

3ObservationsofIOfromsatellite

Figure3.4:MapoftheSouthernHemi-
sphereshowingthecoastlinesoftheAntarc-
ticcontinent.ThelocationofHalleyRe-
searchStation(75.5◦S,26.5◦W)notfarfrom
theWeddellSeaismarkedbythebluecircle.

Figure3.5:TimeseriesofSciamachyobservationsaboveHalleyResearchStation,Antarctica.
AllHalleysatelliteStation.Themeasuremenbluetsdotsarerepresenconsideredtdailywhicvhfallalues,thewithinblacabkoxsolidwithline500isakmwsideeeklylengthrunningenclosingmean
asabetterguidetotheeye.

sun(reducedSZA)andaresultingsmallerAMFasdiscussedinSec.2.3.However,thiseffectis
comparablysmallandmayonlyexplainachangebyaround15%hereandisnotthecausefor
theobserveddecreasefromaround7×1012molec/cm2inOctoberdownto2×1012molec/cm2in
summer.Thismustbecausedbyotherreasons.
Theinterpretationofthisseasonalcycleneedstotakeintoaccountatleasttwodifferentaspects
-ononehandtheamountsandavailabilityofiodinecontainingprecursorsubstancesandonthe
otherhandthephotochemicalconversionsandotherchemicalpathwaysthatleadtoformationof
IO.DirectsourcesofIOarenotknownuptodate,sofirstofall,someprecursorsubstanceshaveto

100

3.2ObservationsofIOinAntarctica

bepresent.Theirdistributionandtemporalevolutionaroundtheyearwillstronglyinfluencethe
IOamounts.Althoughthesesubstanceshavenotbeenmeasuredyetforanentireyearanywhere
nearAntarctica,severalpiecesofinformationareavailablefromdifferentsources,whichhelpto
interpretthepresentfindings(cp.Sec.3.3).

3.2.3Detailedanalysisinhighertemporalresolution

Seasonalvariationforindividualyears
AftertheexampleofHalleyStationshowedthattheannualcyclereappearsinthesameformevery
year,seasonalmapsshallbeinvestigatedforsingleyears,althoughthespatialscatteroftheresults
becomeshigher.Nevertheless,furtherinformationisobtainedfromthesemaps.
Figure3.6compares16maps,showingthefourseasonsforfoursingle,subsequentyears.The
overalltimeperiodisthesameasconsideredabove.Firstofall,thesimilaritybetweenthedifferent
seasonsofindividualyearsisvisible.Inprinciple,thesameevolutionwithhighestandwidest
spreadIOamountsinspringtime,lowervaluesconfinedtoasmallerregioninsummerandagain
higherbutstrongerscatteredvaluesinautumnisobservedeveryyear.TheregionswheretheIO
valuesarelargestagreequitewellbetweenthefouryears,althoughtheexactspatialpatternsshow
differences.Inallautumnandwintermapstheaforementionedhighernoiseappearsduetoasmaller
t.amoundataSomenoticeabledifferencesareseeninthespringtimemapof2006,wherevaluesaresomewhat
higherthanintheotheryearsandtheenhancedvaluesaresomewhatmorewidespreadaroundthe
Antarcticcoast.Furthermore,theautumnmapin2007exhibitslowervaluesonasmallerarea
thantheaverageautumn.Asnoaccompanyingmeasurementsareavailable,thereasonforthisis
currentlystillunderinvestigation.
Despitetheseexistingvariations,themainconclusionfromthiscomparisonisthesimilarityofthe
regionalandtemporalpatternofIO,whichisrepeatedinacomparablewayineachoftheanalysed
years.Observinganarbitraryexampleofoneofthesemaps,onecaneasilyjudgewhichseason
isdisplayed.Averagingacertaintimeperiodoverseveralyearsthereforeleadstoarepresentative
resultfortherespectivetimeofyear.

Resultsinhighertemporalresolution
Monthlyplotsaveragedoverfouryearseachareanalysedinthefollowing.Figure3.7containsin
total16mapsfordifferenttimesofyear.Theaveragingperiodsarespecificallychosensuchthat
subsequentmapsoverlapbyhalfamonth.Thestartdayisgivenintheindividualheadersofthe
maps,thesecondmaphencecontainsdatafortheperiodof15thSeptemberto14thOctober,for
theyears2004-2007.Thisway,arunningaverageovertheIOslantcolumnsisobtained.More
detailedevolutionsnowbecomevisible.
Startinginearlyspringtime,theSeptembermapshowsenhancedvaluesofIOaroundthe
Antarctic,especiallyintheWeddellSeaarea,theshelficeregions(theFilchner-Ronne,theRoss
andtheAmeryiceshelves)andaroundtheAntarcticpeninsula.Ontheoppositesidebetween90◦
and180◦EastandofftheAntarcticcoast,onlysmallnotionsofIOaredetectedbutnowidespread

101

ofationsObserv3

June

-

August

IO

satellitefrom

rebSeptem

-

November

rebDecem

-

ebruaryF

yearFigureforan3.6:overallSeasonaltimemapsspanofofthefourIOyslanearstfromcolumnJuneonthe2004-MSouthernay2008.Hemisphere

102

March

forseparately

-

yaM

heac

(1)

(5)

(9)

(13)

(2)

(6)

(10)

(14)

(3)

(7)

(11)

(15)

3.2ObservationsofIOinAntarctica

(4)

(8)

(12)

(16)

overFigurefour3.7:subsequenMonthlytyearsmapseacofh,theIOstartingslanontthecolumndategivresultseninonthetheindividualSouthernheaders.Hemisphereaveraged

103

3ObservationsofIOfromsatellite

amounts.Asintheseasonalaverage,theslantcolumnsarealsoenhancedontheAntarcticcontinent.
ClosertotheSouthpolethenoisebecomeshigherduetothelowerpositionoftheSun.Goingone
halfmonthfurther,theIOvaluessustaintheirpatternandthemagnitudegrows,e.g.,intheWeddell
Seafromaround6to7×1012molec/cm2.TowardsOctober,thehighestvaluesremainattheirlevels
frombefore,buttheregionofenhancedIOiscontractedclosertotheAntarcticcontinentanddoes
notreachasfaroffthecoastasintheweeksbefore.Exceptforthecontinentandtheshelfice
regions,nowthedirectcoastlinesEastoftheWeddellSeaandtheRossshelficeregionsshowhigh
ts.amounIOAcompletelynewfeaturethenappearsinmap(4)frommidOctobertomidNovember,where
approximately10◦NorthoftheWeddellSeaanextendedregionalbandofenhancedIOdevelops.
Thisbandhasalengthcorrespondingto70◦inlongitudearoundtheAntarcticarea,butisentirely
disconnectedfromtheenhancedIOvaluesclosetothecontinent.InNovember,thenovelcircular
IObandextendsevenfurtheraroundtheAntarctic,thevaluesgrowandthewidthoftheregion
ofenhancedIOisalsoincreased.ItnowreachesfromtheAntarcticpeninsulaatabout60◦West
inEastwarddirectionnearlyuptotheRossshelficeregionat180◦Eastandliesbetween70◦and
60◦South.Thissuddenappearanceofwidespread,enhancedIOamountswithinafewdaysor
weeksonlystronglysuggeststhepresenceofaveryefficientandfastreleasemechanismofiodine
compoundsintherespectivearea.Furtherscientificanalysisaboutpossiblesourcesofiodinespecies
inthisareaaswellascomparisonswithbromineoxidemeasurementsandicecovermapsaresubject
ofthefollowingcentralsectionofthisstudy.
DuringthefollowingweeksuptotheaveragefrommidDecembertomidJanuary(mapno.8),
thecircularstructureissustainedbutchangesitsexactpatternfromweektoweekextendingthe
bandfurtheraroundtheRossSearegionandnearlyclosingthecircleinmapno.7(December
average).Itisremarkable,thattheenhancedIOvaluesovertheshelficeregionsandoverthe
continentdecreaseandactuallyvanishuptoDecemberwhiletheringlikestructureisdeveloped.
Thisindependencyofthetworegionsarguesfordifferentconditionsanddifferentongoingprocesses
-maybeevendifferentiodinesources-ineachcase.
InJanuary,eventuallytheIObandnearlyvanishes,onlysmallremnantsatsomecoastlinesare
stillvisible,whichtwoweekslaterhavedisappearedalso.StartingfromJanuaryagain,IOamounts
reappearovertheshelficeregions(esp.theRossshelficeandattheWeddellSea)andthecontinent
closeby.Theregionexpandstowardsautumnandsimilartoearlyspring,enhancedIOvaluesclose
toandovertheAntarcticcontinentareobserved.
FromMarchonwards,similarasinSeptember,astrongerspatialnoiseandscatterintheabsolute
IOamountsisapparent.Thisisinbothcasesduetolessmeasurementsatthesetimesascompared
totimesclosertoSouthernsummer.InAprilfinally,nosystematicallyenhancedIOamountscan
bedistinguishedfromtheoverallnoiselevelandtheregionispartlyaffectedbywinterdarkness
already,sothatnofurthermeasurementsarepossibleintheSouthernlatitudes.

ThesenovelobservationsofIOaroundAntarcticawithamoredetailedviewonthetemporal
evolutionoftheabundancesinthedifferentareasprovideasubstantialbasisfortheinterpretation
ofongoingiodinereleaseandchemistryatSouthernhighlatitudes.ThemapsofFig.3.7willbeof
valuablehelpforthefollowinginterpretations.

104

3.3DiscussionofobservationsinAntarctica

3.3DiscussionofobservationsinAntarctica

TheobservationsofiodinemonoxideoverAntarcticaarestartingpointsforinterpretationsand
discussionsofpotentialsourcesand(chemical)pathwaysofiodinespecies.Therelevantquestionis
whichprocessesandsequencesmayleadtothedetectedamountsofIOandtotheobservedseasonal
pattern.spatialandcycleAfteriodinespecieshavebeenreleased,chemicalconversionsbetweenthedifferentiodinecom-
poundstakeplace.Inestimatingwhichprocessesmayberelevant,therespectivelifetimesofthe
species,theknownreactionkineticsandthesurroundingconditions,e.g.,otherspecies,thepho-
tolyticsituationsuchastheSZAandthetimeofyear,aswellasthephysicalenvironment(e.g.
thepresenticecover)mayplayarole.Alltheseaspectsinfluencethepathwaysofiodinechemistry.
Thecombinationofthemostrelevantfactorsmustbeabletoexplaintheannualcycleandalsothe
spatialvariationwhichhavebeenobserved.
Consideringthespatialdistribution,twooreventhreedifferentregionsmighthavetobedistin-
guished.ThecircularbandofenhancedIOappearingaroundmidOctober/beginningofNovember
aroundAntarcticamayhavedifferentoriginthantheIOamountsontheshelficeregionsincertain
timesoftheyear.Thehighamountsonthecontinentinearlyspring(SeptembertoOctober)again
tion.attenseparateneed

releaseOrganic

Concerningthesourcesofiodinespecies,therearegenerallytwodifferenttypesofreleaseprocesses,
theorganicandtheinorganicrelease.Iodinereleasehasbeenassociatedwithorganicprecursors
suchasmethyliodideordiiodomethaneandothers(Alickeetal.,1999).Thesesubstancesare
ofbiogenicorigin,e.g.,emittedfromcertainalgaetypes(ReifenhäuserandHeuman,1992).This
connectionhasdirectlybeenobservedinthelaboratory,butalsoinmarinesitesinmidlatitudes.
FortheSouthernOcean,thereleaseprocesseshavenotbeenclarifiedyet.Noparallelmeasurements
ofIOintheatmosphereandpotentialprecursorsubstancesintheairaswellasinthesurrounding
oceanoricesheetshavebeenundertaken.
Inanalogytotheobservationsofthemidlatitudes,itiswellpossible,thatorganicreleaseplays
animportantrolealsointheAntarctic.TheSouthernOceanisbiologicallyveryproductiveand
representsahabitatformanylife-forms.Intheocean,highconcentrationsofphytoplanktontypes
havebeenfound.Thecoldandnutrientrichwaterisespeciallyadvantageousforcertainspecies
suchasdiatoms(ThomasandDiekmann,2003).Theseorganismsevenlivepartlyunderneaththe
icesheetsandtakeadvantageoftheshelterandholdwhichisprovidedthere.Therefore,Antarctic
icesheetshaveoftenbeenreportedtobeofgreenishcolourontheirundersurface(Thomasand
Diekmann,2003).Thereisnodoubt,thatbiologicalprocessesaretakingplaceintheAntarctic
region.Therefore,biogenicreleaseoforganoiodineshastobeconsideredasonepotentialsourceof
IOintheAntarcticatmosphere.
IodinatedmethanespecieshavebeenobservedonshipbornecampaignsintheSouthernOcean
closetotheAntarcticcoast.Alongsidethedetectionofthespeciesinwater,alsomeasurements
intheboundarylayerabovewereconductedby(Carpenteretal.,2007)andsubstantialamounts

105

3ObservationsofIOfromsatellite

of,e.g.,CH2I2weredetected.Asthesewerespotmeasurements(takeninAntarcticsummer),the
situationonalargerspatialandtemporalscaleisnotknown,buttheexistenceofiodinatedspecies
inthemeltwaterclosetothecoastalicesheetsandintheatmospherehasbeenproven.
Ifthephytoplanktonspeciesareclosetotheoceansurface,theemissioncanhappenmore
orlessdirectly.Incaseofanicecoveredarea,therearetwodifferentpossibilities.Seaiceis
oftenofporiferousstructurewithenclosuresofliquidwater(ThomasandDiekmann,2003),so
thatsubstancesmaytravelverticallythroughtheiceifitisnottoothick.Ontheotherhand,
therearealwaysopenleadsandgapsintheseaicecoverwhichmakethedirectreleaseinbetween
theicecoveredareaspossible.Soevenifanicesheetseemstobeclosedwhenobservedfrom
space,theremaybeopenspacesallowingdirectcontactoftheoceanwaterwiththeboundary
layerabove.Furthermore,largeopenleads(polynyas)causeadirectcontactbetweentherelatively
warmoceanwaterandthecolderairabove,sothatpolynyasaretypicallyaccompaniedbyair
convection.Concentrationgradientsacrosstheoceansurfaceareinfluencedbytheconvective
removalofthegaseousspecies.Thisway,thereleaseofgaseouscompoundsfromthewatertothe
airisfacilitated.Throughconvectivemotion,upliftofiodinespecieschangesitsverticaldistribution,
andadditionally,horizontaltransportprocessesmaybethusinitiated.
Consideringthisbackgroundknowledge,theIOalongcoastlinesandabovethecomparablythinsea
icemayindicatebiogenicsources.Icealgaeanddiatomsareregardedthemostprobablecandidates
here,butalsoothersourcesshouldbekeptinmind.Somefurtherideasaredevelopedbelow.

releaserganicIno

Ontheotherhand,alsoinorganicpathwaysmaybeimportantforatmosphericiodinecontentin
Antarctica.Assomeparallelsbetweenthedifferenthalogenspeciesexistingeneral,theknowledge
onthefieldofatmosphericbrominecangivemoreinsighttowhatmighthappeninthecaseof
dine.ioBromineoxide(BrO)canbemeasuredwiththeDOAStechniqueintheultravioletwavelength
range.Forbrominespeciesalsoboth,organicandinorganicemissionprocesses,areindiscussion.
However,forthepolarregions,strongevidencehasbeenfound,thattheinorganicpathwaysareof
majorimportancehere.Severalfieldandlaboratorystudiesleadtotheconclusionthatbiogenic
releasewouldnotbeabletoaccountfortheobservedamountsofBrOandtherapidO3destruction
duringozonedepletionevents.Thephotolysisoforganicbrominespeciesistypicallytooslowto
explainthesuddenandstrongoccurrenceofreactivebromineinthegasphase(Sec.1.4.2).
ThesituationforIOmightbedifferentfromthatofBrO,eventhoughsimilaritiesinthespatialand
temporalpatternshaveraisedtheassumption,thatthereleaseprocessesmaybeconnectedinsome
cases.Asimilaritybetweenthechemistryandsourcesoftwohalogenoxideshasbeensuggested
previouslyfollowingground-basedmeasurementsatHalleyStationbySaiz-Lopezetal.(2007b).
AnadditionalfactmakesthecomparisonbetweenIOandBrOobservationsevenmoreimportant.
Evenifthereleaseitselfturnsouttobedifferentforiodine,thechemistryofbothhalogensissurely
connected,ascrossreactionstakeplaceandthepresenceofiodinemightevenleadtoenhanced
releaseofbromine(Vogtetal.,1999).Therefore,thecomparisonofobservationsmayhelpinany
caseintheinterpretationoftheongoingprocesses.Forthispurpose,themapsofIOandBrOwill

106

3.3DiscussionofobservationsinAntarctica

becomparedinthenextparagraph.Inorganicreleasehasitsbeginningwithintheoceanwater,
wherevariousiodineandbrominespeciessuchasiodide(I−),bromide(Br−),iodate(IO3−),HOBr,
HOIandseveralothercompoundsarepresent.Directinteractionontheoceansurfaceleadsto
theexchangebetweensubstancesintheaqueousandthegaseousphase,suchasI2(aq)↔I2(g).
Additionally,heterogeneousreactionsontheoceansurfacemaytakeplaceinvolvinggasphaseand
aqueousphasespeciesatthesametime.Apartfromtheoceansurface,seasaltaerosols,icesurfaces,
frostflowersandbrinearepossiblelocationsforrelevantinteractionsandmulti-phasereactions.
Locationswhereinorganicreleasemaybecomeimportantincludetheiceedgeoftheocean,fresh
seaice,butalsoallregionsthatcanbereachedbywind-blownaerosolcomingfromtheoceanand
seaice.Especiallyinterestinginthisrespectistheobservationthatseasaltaerosolistypically
enrichediniodineascomparedtoseawatercontent(Murphyetal.,1997).

3.3.1BrOobservationsandiceconcentrationinAntarctica
InordertocomparetheresultsofIOandBrOobservationsfromsatellitewitheachother,Fig.3.8
shows16mapsoftheBrOverticalcolumnsfortheequivalenttimeperiodsasinFig.3.7fortheIO
slantcolumns.TheBrOdatawereprovidedbyMathiasBegoinandAndreasRichter,University
ofBremen(Richteretal.,1998).Intheseverticalcolumns,thestratosphericpartofBrOisstill
contained(Sinnhuberetal.,2005;Rozanovetal.,2005a).Thisisnotnegligible,butspatiallyslowly
varying.ThestratosphericportioniscoveredbythegreenvaluesfromthecolourscaleforBrO
startingat4×1013molec/cm2andnotatzero.Thestrongervaryingtroposphericportionappears
inyellowtored.Theconversionfromtheoriginalslantcolumnstothedepictedverticalcolumns
wasperformedusingastratosphericAMFatanalbedoof90%andindividualSZAandwavelength
values.TheAMFisvalidforthestratosphericportionofthecolumnandslightlyunderestimates
thetroposphericamountsofBrO(Richteretal.,1998).
AlthoughthemapsofIOshowslantandnotverticalcolumns,acomparisonisstillmeaningful,as
nottheabsolutevaluesbutmorethespatialdistributionandtemporalvariationsareofinterestfor
theinterpretationhere.Withoutknowingtheindividualprofilesofthetracegasesandtheground
spectralreflectances,theconversionbetweentheSCandtheVCforIOhereisgivenbyafactor,
whichchangesonlyslightlythroughouttheyeardependingontheSZA.
ForthecaseofBrO,theseaiceconcentrationisknowntobeofmajorimportanceforthe
interpretationofBrOobservationsinPolarRegions(Kaleschkeetal.,2004).Fromthecurrentnew
investigationsinthisstudy,itbecomesprobable,thatalsoIOisincertaincasesrelatedtothesea
er.vcoiceInordertocomparethehalogenoxidemapstoseaicecover,Fig.3.9showsmapsoftheice
concentrationinandaroundAntarcticafromtheAMSR-Esensor(cp.Sec.1.9.3),providedby
GunnarSpreenandLarsKaleschke,UniversityofHamburg(Spreenetal.,2008).Inthisplot,also
16mapsaredisplayed,buteachmapshowstheicecoversituationforonecertaindaygivenat
thetopofeachmap.ThechosentimeisSeptember2006toMay2007andeachdayinthisfigure
correspondstothemiddleofthemonthlyaveragingperiodfromFigs.3.7and3.8.Thesequence
oficecovermapsthereforestartsonthe15thofSeptember,whiletheIOandBrOmapsshowfirst
anaveragefrom1stto30thSeptember.Eachfollowingmapineachsequencewastakentwoweeks

107

3ObservationsofIOfromsatellite

(1)

(5)

(9)

(13)

(2)

(6)

(10)

(14)

(3)

(7)

(11)

(15)

(4)

(8)

(12)

(16)

Figure3.8:MonthlymapsofBrOverticalcolumnsforthesameaveragingperiodsasforIOin
3.7.Fig.

108

3.3DiscussionofobservationsinAntarctica

later.Theexacticecovernaturallyvariesfromyeartoyear,sothatthespatialpatternexhibitsdeviations
whencomparingonespecifictimeperiodwiththesameperiodofadifferentyear.However,the
seasonalchangesarestilllargerthantheyear-to-yearvariations,andatypicalicecovermaybe
allocatedtoacertaintimeofyear.Forthegeneralcomparisonhere,theexactpatternisnot
relevant,butmorethegeneralevolution.Asthegeneralpatternsareroughlyrepeatedannually,
thecomparisoncanbeperformedusingdailymaps.
BrOcolumnsarealreadyabovethedetectionlimitinAugust(notshownhere),andawell
pronouncedcircularpatternofenhancedBrOamountsaroundAntarcticaisseeninSeptember
(Fig.3.8,mapno.1).BycomparingtheBrOandicecovermaps,theknownobservationcanbe
reproduced,thattheenhancedBrOcolumnscorrelatespatiallywiththeextentinseaicecover
(Kaleschkeetal.,2004).ThecircularstructureintheBrOcolumnsremainsnearlyunchangedwith
onlyslightvariationsuntilmid-Octobertomid-November(map4).Frommap5onwards,additional
regionswithhighBrOvaluesappear-theRossSeaaround180◦Eastaswellastheshelficearea
andsomecontinentalpartsofAntarcticaclosetotheWeddellSeaandfurtherSouthwards.Also,
highBrOvaluesaroundthecoastlinesofAntarcticabecomevisible.ThisenhancementofBrO
closetothecoastremainspresentthroughoutthesummertimeuntilMarch(map13).Thebroad
ringofenhancedBrOhowevervanishesaroundDecember(maps7-8).Fromthistime,theshelf
iceregions,somecontinentalareasclosebyandthecoastlinesarethemainregionsofhighBrO
amounts.Frommid-Marchonwards,hardlyanyBrOcanbedetectedontheSouthernHemisphere
erages.vatheseinConcerningthecomparisontotheseaicecover,thelargestareasofhighBrOarefoundabove
theextendedseaicesheets.UptotheendofFebruary,thepatternofenhancedBrOfollowsquite
wellthespatialextentoftheseaicecover.Oneexceptionaretheshelficeregions,whereBrOis
presentforsomemonthswithoutdirectcontacttoseaice,e.g.inmaps4-8,fortheshelficeregions
oftheRossSeaandtheWeddellSea.Inautumn,theBrOdoesnotcorrelatewiththeseaicecover
whenthisbeginstogrowagain,buttheBrOremainsbelowdetectionlimit(maps14-16).
ThedetectedcorrelationsbetweenBrOandseaicecoverandthehighamountsofBrOhave
leadtotheassumption,thatBrOismostprobableofinorganicoriginfromefficientprocesseswhich
takeplaceonsaltysurfacesandundercertainconditions(Sanderetal.,2006a;Simpsonetal.,
2007a).Theconceptofpotentialfrostflowersandtheconnectiontobrineonseaiceweredeveloped
(Kaleschkeetal.,2004).Uptonow,itisnotclarifiedwhichactualprocesses,conditionsorspecific
locationsontheseaiceareresponsibleforthebrominerelease.Itisagreedthatprobablytheso-
calledbromineexplosiontakesplace(cp.Sec.1.4.2).Thisauto-catalyticreleasemechanismleads
tofastandefficientreleaseofbromineatomstotheatmosphere,directlyfromtheseasaltcontent
withoutinvolvementoforganicprocesses.
ForthespatialpatternofBrO,alsotransportplaysanimportantrole.Theenhancedvalues
ontheshelficeregionsforwhichnolocalreleaseprocessesareproposed,arecausedbytransport
processes.BrOmaybecarriedoverfairlylargedistancesontheorderofseveralhundredkilometres
withinseveralhoursuptoafewdays(Begoinetal.,2009).

109

3ObservationsofIOfromsatellite

15.09.2006

15.11.2006

15.01.2007

15.03.2007

01.10.2006

01.12.2006

01.02.2007

01.04.2007

15.10.2006

15.12.2006

15.02.2007

15.04.2007

01.11.2006

01.01.2007

01.03.2007

01.05.2007

Figure3.9:DailymapsoficeconcentrationontheSouthernHemisphereretrievedfromAMSR-E
measurementsinthe89GHzchannelataresolutionof6.25km(Spreenetal.,2008).Themaps
wereobtainedfromGunnarSpreenandLarsKaleschke,InstituteofOceanographyattheUniversity
ofHamburg,GermanyinJune2009(digitalmediaftp-projects.zmaw.de/seaice).Thedataversion
is”AMSR-EASI6.25kmSeaIceConcentrationData,V5.5i”.

110

3.3DiscussionofobservationsinAntarctica

3.3.2ComparisonofIOwithBrOandseaicemaps
OnesimilaritybetweenIOandBrOobservationsinAntarcticaisfirstofallthegeneraloverlapin
timeandspaceofenhancedcolumns.ThelargestamountsofIOgloballyarefoundinthespring
timeAntarctic.ThesameholdstrueforBrO,althoughcolumnsofcomparablemagnitudearefound
intheArcticalso.Severalsimilaritiesbetweenthetwospeciesareobserved:
•Thecircularstructurewhichappearsinboth,theIOandtheBrOmaps,isofsimilarpattern
andlocationapproximatelyfrommid-OctobertoendofDecember(maps4-8).Bothhalogen
oxidesshowaconnectiontoseaiceconcentrationduringthisperiodwithhighestamounts
appearingabovethepresentseaice.
•Secondly,theenhancementovertheRossSeaandtheWeddellSeashelficeregionsoccursfor
bothspecies.However,thetimeswhenhighvaluesaredetectedarenotexactlythesame.
•AlthoughthisisnotasprominentforIOasforBrO,bothtracegasesshowenhancedamounts
alongthecoastlines.WhileBrOisdetectedaroundthecoastfornearlyallperiodsshown,
thevaluesdropbelowdetectionlimitforIOaroundendofJanuaryandbeginofFebruary
(maps10and11).Nevertheless,smallerpatchesofIOamountsaroundthecoastareseenin
maps.theofmostOntheotherhand,thepatternsofIOandBrOalsoexhibitsomeremarkabledifferences:

•IOvaluesareenhancedabovethecontinentinSeptemberandOctober,whilenoBrOappears
abovethecontinentatthattimes.TheBrOisdistributedatafurtherdistanceinthedescribed
circularpattern,whereasIOamountsaremoreconfinedtowardsAntarcticaanddonotreach
asfarontotheseaiceinSeptemberandOctoberaslateron.TheconnectionbetweenBrO
andseaiceconcentrationseemstobestrongerandvalidforalongerperiodoftimethanfor
IO.•Concerningtheshelficeregions,bothspeciesaredetectedhere,buthighIOamountsareseen
nearlythroughouttheentiretimeserieswithlowervaluesaroundDecember(map7),while
BrOshowsthehighestamountsintheshelficeregionsbetweenNovemberandJanuary(maps
5-9).•InMarch,IOvaluesarenearlyaswidespreadasinSeptember,butBrOamountsmostlydrop
belowthedetectionlimitexceptforsmallindicationsaroundthecoastwhichremainuntil
h.Marcofend

3.3.3IOinseaicecoveredareas
Thecases,wherecorrelationsbetweenIOandBrOobservationsaredetectedsuggest,thatsimilar
orconnectedreleasetakesplace.Thismayespeciallybetrueforthehalogenoxidesdetectedon
seaiceandalongthecoastlinesofAntarctica.Inthecentralreactionoftheinorganicbromine
explosion,twoinactivebrominespeciesconverttoBr2withinseawaterorseasaltaerosol:
HOBr+Br−+H+→H2O+Br2(R16)

111

3ObservationsofIOfromsatellite

ThismayleadtoanexponentialincreaseintheconcentrationofBrO.Oneimportantprereq-
uisiteforthistohappenisthenecessaryacidity(involvementofH+ions).Inthiscentralreaction,
bromidemaybeexchangedforiodideandthenalsoleadtocompletelyinorganicreleaseofiodine
ecies:spHOBr+I−+H+→H2O+IBr(R39)
Inthiscase,IBrissetfreewhichyieldsaniodineandabromineatomafterphotolysis.The
abovecrosshalogenreactionshowsoneoftheexistinglinksbetweenbromineandiodinechemistry.
AnotherimportantreactionisthereactionofIOwithBrO,whichalsoyieldsIBrinthegasphase.
ThisisespeciallyrelevantfortherecyclingofBratomsastheselfreactionofBrOprobablydoes
nottakeplace.Thisconsiderationsuggestsastrongcorrelationbetweenthetwospecies,asmutual
amplificationoftheirreleaseandrecyclingprocessesispossible.
ThecircularpatternofBrOaroundAntarctica,however,occursalreadyearlierintheyearthan
forIO.Thisarguesforconsiderabledifferencesinthereleaseprocess.Possiblytheiodinereactions
needdifferentconditionsthanthebrominereactionswhicharegivenonlylaterintheyear.Asecond
possibilityistheinvolvementofbiologyatthispoint.Whilethechemistryofiodineandbromine
issurelyrelated,theactivationmightstillundergodifferentpathways.Asstatedabove,seaice
isusuallynotaclosedsurfacebutexhibitscracksandlargeopenleads(ThomasandDiekmann,
2003).Asphytoplanktonispresentunderneaththeseaicesheets,emissionofbiogenicsubstances
maybeenhancedinthepresenceofopenareas,brokenicesheetsandevenjustthinnericesheets.
TheIOappearsontheseaicearoundlateOctober,althoughsunlightenterstheseareasseveral
weeksearlierwherenoIOisdetectedonthedistantseaice.
Presumably,theiodinereleaserequirestheseaicetobemoreporousandinstablethanitisin
earlyspringtime.Fromthebeginningofspringtowardsthesummer,theseaicegetsthinnerandis
moreofteninterruptedbyopenwaterleadsandpolynyas.Thiscanbefollowedintheicemapsof
Fig.3.9,wheregreyishcoloursmarkreducediceconcentration.Inthissituation,possiblybiological
activitymightbecomemoreprominent.
DiscussingthepotentialbiologicalsourceofhalogensintheAntarcticitmaybeilluminating
toconsiderthebiologicalactivityintheoceans.Fromspace,thiscanonlybemeasuredinicefree
areas.Apotentialindicatorofactivebiologyintheoceansisgivenbythechlorophyllconcentrations.
Chlorophyll,theinitiatorofphotosynthesis,isagreenpigment(absorbinginblueandredspectral
bands)whichispresentinplants,andalsoinalgaeandcyanobacteria.Figure3.10showsthe
SeaWIFSmonthlyclimatologiesofchlorophyllaconcentrations(cp.Sec.1.9.3)forthesixmonths
ofOctobertoMarch(Hookeretal.,1992).Eachmapcontainsdatafrom10years.Thechlorophyll
a(Chla)concentrationiscolourcodedasspecifiedbythecolourbarinthebottomofthefigure.
Regionswithoutdataappearblackinthereadilyprovidedmaps.Withagroundresolutionof9km
sidelength,oceandataisonlyavailableforicefreeareasofthissize.Theradiationreflectedfrom
icesheetsinthefieldofviewotherwiseoverbalancestheoceansignal.Intheremainingmonthsnot
shownhere,theoceansurroundingtheAntarcticismostlyicecoveredandonlysmalleropenleads
andpolynyaswithlessthan9kmsizemayexist.

112

erOctob

rebDecem

ebruaryF

3.3DiscussionofobservationsinAntarctica

November

uaryJan

hMarc

withFigure9km3.10:spatialSeaWIFSresolutionmonbythlyNASAclimatologies(NASA,ofhthettp://ooceanicchlorophceancolor.gsfc.nasa.goyllaconcenv,Junetration,2009).provided

113

3ObservationsofIOfromsatellite

IntheSeaWIFSimages,thedevelopmentofopenwaterregionswithintheseaicecoverin
Antarcticabecomesvisibleascolouredregionsappearingintheblackenedspace.Itisinterestingto
notethattheseopenwaterregions,especiallyinJanuaryandDecemberbutalreadyinNovember,
showenhancedchlorophyllconcentrationsindicatingstrongbiologicalactivityintheseareas.In
November,especiallytheoceanclosetotheRossSeaexhibitsenhancedchlorophyllamounts.In
DecemberandJanuary,alsoenhancedchlorophyllintheWeddellSeaareacanbeclearlydistin-
guished.Theamountsarecomparablylarge.Thesemapsdonotgiveinformationontherespective
specieswhichareresponsibleforthechlorophyllconcentrations.Ingeneral,specieswhichrelease
iodineprecursorsaswellasthosewhichdonotinfluencetheatmospherichalogenlevelscontribute
tothechlorophyllasignal.However,themapsprovethatbiologicalactivityishighinmanyloca-
tionsassoonastheiceopens.Itseemslikelythatbiologicalactivityalsooccursinsmalleropenings
intheicewhicharenotresolvedintheseimages,butwhicharedefinitelypresent(Thomasand
Diekmann,2003).Formoredetailedanalysisonasmallerscale,e.g.,intheseaicecoveredregions
stillinOctoberandevenearlierinSeptember,datainhigherresolutionneedstobeconsideredin
thefuture.Themostimportantoutcomeoftheseimagesistheexistenceofhighchlorophyllcon-
centrationsclosetoAntarcticaintimes,wherealsoIOhasbeenobserved.Theseimagestherefore
supporttheidea,thatorganicsourcesofiodineprecursorsmayatleastpartlycauseandinfluence
thehighiodineoxidecontentoftheAntarcticatmosphereincertaintimes.

3.3.4IOonAntarcticshelficeregionsandthecontinent
UnexplainedsofararethehighIOamountsontheAntarcticcontinentinearlyspringtime(i.e.
SeptembertoOctober).TheoriginalexpectationwastofindIOaroundAntarctica,similartoBrO,
stillinthevicinityoftheocean,butnotonthecontinent.Theseresultmayneedfurtherinvesti-
gationsinthefuture.NodirectsourcesinlandonAntarcticaareknownuptonow.Nevertheless,
someexistingprocesseswouldleadtohighIOamountsonthecontinentalso,andmayexplainthe
observations.Twoimportantaspectsneedtobebroughtforward.

I.TransportprocessesofIOandiodinerichaerosols
Firstofall,transportprocessesfromotherregionsneedtobetakenintoaccount.AfterIOhas
formedintheatmosphere,itcanbephotolysedquitequicklybacktoiodineatoms.Therefore,its
individuallifetimeisshortandonemightrashlyassumethatlongrangetransportconsequentlyisnot
relevant.However,moredetailedconsiderationspointoutthepossibleoccurrenceandsignificance
nonetheless.orttranspofIOandBrOareoftendetectedontheiceshelves.Similartothesituationonthecontinent
furtherinland,nodirectsourcesofbromineoriodinehavebeenreportedfortheshelficeregions.
Interestingly,transportprocessesreachingfarinlandcanberegularlyobservedforthecaseofBrO
(Begoinetal.,2009).BrOcolumnshavelargeropticaldepthsandcanbeanalysedonadailybasis,
whichisnotpossibleforIOasvaluesaremuchclosertothedetectionlimit.Therefore,individual
transporteventscannoteasilybedetectedinthesatelliteIOdata.Consideringtheobservation,
thatBrOistransportedoverthousandkilometresinland,thismayalsobetrueforIO.Thetimes
involvedinsuchtransportprocessesaremuchlongerthanthetypicalphotolyticlifetimesofBrO.

114

3.3DiscussionofobservationsinAntarctica

Thelifetimesofthehalogenoxideshavealargeinfluenceonthepossiblepathways.Andsome
importantconsiderationsshouldbemade.BrOlifetimesduringstrongexposuretosunlightlie
aroundtwominutes,forIOaroundseveralseconds.Thisbasiclifetimeismuchshorterthanthe
effectivelyrelevantlifetime(Simpsonetal.,2007b),astheBrandIatomsproducedbyphotolysis
reactwithO3toformbackBrOandIO.SotheBrOandIOarenotlostbyphotolysis.Theeffective
lifetimeissubstantiallylongerandforthecaseofBrOx(BrOx=BrO+Br),lifetimesaroundmany
hourswereestimatedbySimpsonetal.(2007b).
TheequivalentestimationwasnotperformedforIO,butasimilarconsiderationholdsforreactive
iodine.Forsimilarsituations,thelifetimeofIOiseffectivelyenhancedbythecyclingofreactive
iodinebetweenIandIO,sothatthelifetimeofIOx(IOx=IO+I)issubstantiallylargerthanfor
molecule.IOindividualanInadditiontotheenhancementoftheeffectivelifetime,furtherprocessesmaytakeplacewhich
additionallysustainhighlevelsofhalogenoxidesandenabletransportprocesses.Concentrating
onthecaseofIO,onespecialcharacteristichassubstantialinfluence.IOformshigheroxidesina
selfreaction,firstgeneratingOIOmolecules,andthereafteralsohigheroxidesofthegeneralform
IxOy(cp.Sec.1.5.3).Iodineoxidesleadtotheformationoffineparticlesthemselvesandtheyalso
undergoheterogeneousphasereactionsonthesurfacesofaerosols,snowandice.Thisisfacilitated
bythethehygroscopicityofthesegases.Beingattachedtotheaerosolphase,theiodineatoms
involvedarenotconsumedbutmaybetransportedoverlongerdistancesthanpossiblesolelyin
thegasphase.Lateron,theiodinespeciesmaybereleasedfromtheaerosolphaseagainandthe
reactiveiodineisrecycled.Thechemicalreactionsofthisrecyclingarenotknownyet,butasthe
condensableiodinevapoursandfollowingparticlesarehygroscopic,somereactionsinthefollowing
aqueousphaseorontheaerosolsurfacemayproceed.Dependingontheavailableiodinecompounds
inthesolution,someiodinemoleculesmaytransferbacktothegasphase,bephotolysedandrecover
someIO.Thisway,thereactiveiodinecaneffectivelytravellongerdistancesthaninitiallyassumed.
Allpossibleconversionsandpathwaysneedtobeconsideredwhenestimatingtheeffectivelifetime
ofIO,andtransporttoregions,wheredirectsourcesarenotknownmaywellbepossible.The
significanceofthecombinedrecyclingandtransportprocessesneedstobeevaluatedmorecarefully
inordertoreceivemeaningfulnumbersofthepossibleamountsofIOtransportedacertaindistance
tarctica.Anofinland

II.Unfamiliarandunusualsources
NoinlandsourcesofiodineinAntarcticaareknownuptonow.However,thisdoesnotprovethat
suchsourcesindeeddonotexistandformerlyunknownornotconsideredorevenforgottensources
needtobe(re)considered.Someideasandproposalsofprobablyminorimportanceshallbeshortly
mentionedhere,ofwhichmanyarestillopenforcloserinspection,buttheyshouldnotbeexcluded
beforetheirimportanceisproperlyassessed.
SeveralvolcanoesarepresentinAntarcticasuchasMountErebusat77.5◦S,167.2◦Ecloseto
theRossSea.Volcanicactivityisknowntobeasourceofhalogensandhalogenoxides(Bobrowski
etal.,2003).HowimportantthismightbefortheAntarcticregionissofarnotknown,butamajor
importanceseemsimprobableasvolcanoesinotherlocationhavenodetectableinfluenceontheIO

115

3ObservationsofIOfromsatellite

amounts,andtheactivityofAntarcticvolcanoeshasnotbeendistinguishedasespeciallystrong.
AlsonodirectevidenceforBrOfromtheAntarcticvolcanoeshasbeenobserved.
Futhermore,salt-richdrylakesarecommoninsomeAntarcticlocations.Thesaltcontentof
thesedrylakesisextremelyhigh,forexample,intheMcMurdoDryValleyclosetotheRossice
shelf.Theemissionpathwaysfromtheseplacesneedtobepointoffurtherresearchasthedetails
arenotknownyet.Byerosionorotherprocesses,salt-richaerosolsmayentertheatmosphere.
Extremelysaltysoilswerefoundandtheyareexpectedtobedevelopedbywind-blownseasalt
(Mahaneyetal.,2002).TheabilityofIOtoformhigheroxidesandfineparticlesfurtheronmight
influencethesituationinfavourofiodinecompoundsintheseevents.Thisadditionallysupports
theargumentabove,thattransportprocessesofseasaltaerosolneedtobeconsideredasahalogen
sourcefortheAntarcticcontinent.
Backin1982,Rasmussenetal.havereportedonmethyliodideobserveddirectlyattheSouth
Pole.Mixingratiosof1.8pptCH3Ihavebeendetected.Consequently,someunidentifiedlocal
sourcesofiodinespeciesincontinentalAntarcticaarepresentortransportmustberesponsible.
Furthermore,bacteriamightbeofgreaterimportancethanpreviouslyassumed.Researchefforts
inquantifyingiodineemissionsfrombacteriahaveshownthatbacteriaarecapableofmethylating
iodidewhichisavailableinseawater.Itisconcludedthatbacteriacanstronglycontributetothe
transferofiodinespeciestotheatmosphere(Amachietal.,2001).Thedistributionofbacteriain
theAntarcticregionwouldneedtobeanalysedinthisrespect,asdifferentbacteriahavedifferent
capabilitiesofmethylatingiodide.TheimportanceofbacteriaforIOlevelsinAntarcticacannot
beestimated,butshouldnotbedisregarded.

116

3.4ObservationsofIOintheEasternPacific

3.4ObservationsofIOintheEasternPacific

Intheglobalmaps(Figs.3.1,3.2),someenhancedIOamountsovertheEasternPacificwerepointed
out.Figure3.11showsacloseupoftheEasternPacificregion,takenfromthefouryearaveragedIO
map(cp.Fig.3.1).AlongthePacificcoastofEcuadorandPeruandalsoofNorthernChile,enhanced
IOamountsaround4×1012molec/cm2areidentified.Theenhancedvaluesareconcentratedover
theoceananddonotextendoverland.Thisfindingindicatesthatsomereleaseprocessclosely
linkedtotheoceanicsitetakesplace.
Thisregionbelongstoahighlyinterestingarea,wherebiologicalproductionisstrong.This
willbefurtherexplicatedbelow.However,theanalysisinthisregionismorechallengingthan,
e.g.,fortheAntarctic.AsdiscussedinSec.2.4,detectionlimitsandnoiseinfluencesarestronger
andplayalargerroleinoceanicregionsthanabovehighreflectingsurfaces.TheIOresultsinthe
EastPacificappeartobelessstablewithrespecttochangesintheretrievalsettingsthanforthe
Antarctic.Nevertheless,thecurrentresultsyieldenhancedIOamountsandhintatsomebiogenic
releaseprocessesintheEasternPacificregion.
IntheseasonalmapsofIO,someannualvariationshavebeenvisibleinthisregion.However,
sometemporalvariationsarepresentinawidesurroundingareaandnotonlyatthelocations
whereenhancedamountsaredetected.Astheareaisaffectedonalargescaleandthevariationis
systematicwiththecourseoftheyear,thisvariationmightnotbearesultofactualIOamounts
butratherofasystematicretrievalimpactcausedbygeometricalinfluences.Thepositionofthe
satellitewithrespecttothesunislargelythesameovertheareaforagivenday,butthegeometry
changeswithseason.Dependingonviewinggeometry,therelevantscatteringanglesandwiththis
thetypicalpolarisationdegreewillvaryandpossiblyimpactontheretrieval.
AtimeseriesofIOcolumnshasbeenselectedinaregionfrom10◦Sto5◦Naswellasfrom100◦
Wto80◦W.Inordertoremovethelargescaletemporaldrifts,theaverageintheoverallregioncon-
tainedinFig,3.11hasbeensubtracted.InFigure3.12theresultingIOslantcolumndensityisplot-
tedversustimefortheyears2005to2007,showingdailyamounts(greytriangles)andthemonthly
mean(redlinesandsymbols).Thetypicalstandarddeviationamountsto4×1012molec/cm2,the
errorbarsarenotplottedtothedatapointsforclarity.Theoverallmeanofthetimeseriesamounts
to3.1×1012molec/cm2.TheIOamountsperformonlyslightvariationsthroughouttheyear,the

Figure3.11:IOslantcolumnsover
theEasternPacificregionaveraged
over4yearsfromJune2004toMay
2008.Enhancedamountsarevisible
atlocationsalongthePacificcoast
linesofEcuador,PeruandNorthern
Chile.

117

3ObservationsofIOfromsatellite

Figure3.12:TimeseriesofIOslantcolumnsintheEasternPacific.Fromtheoriginaldataset,
theaverageinawidesurroundingareahasbeensubtracted.Dailyaverages(greytriangles)aswell
asmonthlymeans(redlineandsymbols)areplotted,whilethestandarddeviationsforthedata
pointsliearound4×1012molec/cm2andarenotshownforclarity.Thesmallseasonalvariationis
t.significannot

differencebetweenthesummerandwintermonthsliesaround1×1012molec/cm2.Takinginto
accountthetypicaluncertainties(Sec.2.5)andthestandarddeviationforthedailyaverages,this
seasonalvariationisnotsignificant.Thequestionforaseasonaltrendinthisregionistherefore
presentlyunclarified,butwillneedtobesubjectoffurtherinvestigation.ThefactthatIOamounts
areenhancedalongthecoastline,however,remainsunaffectedfromthisissue.

Connectionstotheoceancurrentandthebiosphere
Themostinterestingconsiderationhereconcernspotentialsourcesofiodinespeciesinthislocation.
TheEasternPacificiswellknownforbeingaso-calledupwellingregion,wherethetypicalocean
circulationisflowingupwards,bringingcoldandeutrophicwatersupfrombelow.Thisisgenerally
leadingtoanenrichmentofsurfacewaterswithnutrients,generatingasuitableenvironmentfor
phytoplanktonandotherspecies.Asaconsequence,suchareasarebiologicallyveryactive.The
upwellingregionhereispartoftheHumboldtcurrent,acoldoceancurrentintheWestofSouth
AmericacomingfromAntarcticaandflowingtowardstheNorthandthenturningWestwards(see
alsoFig.3.13).Itisoneofthelargestupwellingsystemsoftheworldandahighlyproductive
ecosystem.TheconnectionbetweenthebiosphereandiodinespecieswasdiscussedinChapter1and,
accordingtocurrentknowledge,severalalgaeandphytoplanktontypesemitiodinecompounds.So
thenutrientrichupwellingregionsarepotentialsourceregionsofemissionsofiodinecompounds.
Currently,noinformationisavailable,onhowmuchiodineprecursorsareemittedintheEastern
Pacific.Consideringtheabovethoughtsthough,enhancedIOamountsinthisregioncouldwellbe
connectedtothehighlyproductivebiosphere.TheHumboldtcurrentwhichflowspasttheAntarctic
andsurfacesattheSouthAmericanWestcoast,partlyconnectsthebiospheresoftheAntarcticand
theEasternPacific,sothatsimilaralgaeorphytoplanktonspeciesexistinthetworegions.This

118

3.4ObservationsofIOintheEasternPacific

Figure3.13:TheflowoftheHumboldtCurrent
(alsoPeruCurrent).Theexcerptistakenfroma
mapofOceanCurrentsandSeaIcefromtheAt-
lasofWorldMaps,UnitedStatesArmy
ServiceForces.ArmyServiceForces
(1943).M-101Manual

mayformalinkbetweenthehighIOamountsintheAntarcticandtheenhancedamountsoverthe
acific.PEasternAstheEasternPacificisaregionofstrongbiologicalproductivity,thechlorophyllaconcentra-
tionsareexpectedtobehighthere.Toinvestigatethis,aspecificdatasetofSeaWIFSchlorophyll
aconcentrationshasbeenextracted.Figure3.14coversthesametimeperiodandspatialregion
asFig.3.11.ThecolorcodedChlaconcentrationsareclearlyenhancedalongtheSouthAmeri-
canNorthwestcoast,overlappingwiththelocationsofenhancedIOamounts.Biologicalemissions
thereforearemostlikelytheoriginofiodineprecursorcompoundsinthisarea.
ChlafromSeaWiFSJun2004-May2008
Figure3.14:SeaWiFSchlorophyll
aconcentrationintheEasternPa-
cific,averagedfromJune2004to
May2008,i.e.,overthesame
timeperiodasshowninFig.3.11.
Thisimagewasacquiredusingthe
GES-DISCInteractiveOnlineVi-
Infras-aNalysisANdsualizationtructure(Giovanni)aspartofthe
(GES)NASA’sDataGoandddardEarthInformationSciencesSer-
(DISC).terCenvices

Chlaconcentration[mg/m3]

Consideringthepointsofconcernmentionedearlier,e.g,theirregularitiesoverclearwater
regions(Sec.3.1)whereIOcolumnstendtobenegative,andthelowerstabilityofIOresultsover
theocean,somecautionisnecessaryinthediscussionhere.Spectralcorrelationscaninprinciple
impactontheretrievedIOamount.ThequestioniswhethertheapparentlinkofIOtobiologyin
thisareaisaconsequenceofcorrelationsorofcausality.AspectralcorrelationofIOwithchlorophyll
aabsorptionmaybeexcluded,astheglobalpicturesshowthathighChlaconcentrationsdonot

119

3ObservationsofIOfromsatellite

necessarilyimplyhighIOamounts.Forthispurpose,themissioncompositeofSeaWIFScholorphyll
aconcentrationsisshowninFig.3.15containingdatafromAutumn1997toSummer2009.

com-missionSeaWIFS3.15:Figurepo2009siteshofromwingtheAutumnchloroph1997ylltoconcenSummertra-
ceans.otheintion

Severalcoastalregionsarecolouredredduetohighchlorophyllandhencestrongbiological
activity.Theso-calledoceandeserts,e.g.,inthemiddleofthePacificandalsotheotheroceanscan
bedistinguishedbytheirblue/purplecolour.TheareaintheEasternPacific,whereenhancedIOis
seenincertaintimesisnotasstronglypronouncedasotherregions,butexhibitsmediumamounts
ofchlorophyll.Ifaseasonalvariationispresent,amountscanwellbehigherforcertaintimes.The
spatialpatternsofIOandChlaarenotcorrelated,ageneraldependencyofthetwoquantitiescan
beexcluded,thereareseverallocations,whereChlaishighandIOisnot,butotherswhereboth
quantitiesareenhanced,e.g.,asintheEastPacific.
DuetotheproposedlinkofIOamountstotheupwellingregion,thereshouldexistadepen-
dencebetweenthestrengthofIOemissionsandtheconditionswithintheEastPacificwaters,like
temperature,salinityandotherparameters,whichagaininfluencethebiosphere.Consequently,
itisaninterestingtaskforthefuture,toobservepotentialvariationsofIObetweenElNiñoand
nonElNiñoyears.DuringyearsexhibitingtheElNiñoeffect,theHumboldtcurrentconsiderably
weakens,seasurfacetemperaturesareseveraldegreeshigherandbiologicalproductiondecreases.
Inalongertimesseriesofsatellitedata,aneffectonIOmaybecomevisible.
ThequestionariseswhyIOisstrongaboveoneupwellingregionbutnotovertheothersalso,
forexample,attheNorthwestcoastofAfrica(Mauretanianupwellingregion),andalsonotoverall
regionsofstrongbiologicalactivitycharacterisedbyhighChlaconcentrations.Whileinsomeof
theseregions,theIOcolumnsmayjustremainbelowdetectionlimit(cp.Sec.2.4),oneimportant
reasonsforthespatialdifferencesarethespatiallydiversebiospheres.

IOspatialcorrelationwithdiatomconcentration
Ithasbeenfoundthatdifferenttypesofalgaeandphytoplanktonreleasedifferentmixturesand
amountsoforganiccompounds,someorganismsareespeciallystrongemittersforcertainiodocar-
bons(Schalletal.,1994;TokarczykandMoore,1994).Iodinespeciesshouldonlybereleasedover
regionswhicharepopulatedbycertainalgaeandphytoplanktongroups,whichagainshowindivid-
ualspatialdistributionsintheworld’soceans(Alvainetal.,2005;Bracheretal.,2009).Species
suchasdiatomsorheptophytescanbedistinguishedbyopticalmeansbecauseeachspeciescontains

120

60

0

-60

3.4ObservationsofIOintheEasternPacific

differentchromophoresordifferentratiosofcommonchromophoresleadingtocharacteristicspectral
signatures.ThespatialdistributionsofdifferentspeciesmayberetrievedusingtheDOAStechnique
inacomparablylargewavelengthwindowduetoratherbroadabsorptionbands(Bracheretal.,
2009).Specialattentionisdrawntodiatoms,whicharereportedtoemitiodinespecies(Moore
etal.,1996)withspecialactivityofpolardiatomcultures(HillandManley,2009).
Figure3.16comparesthediatomconcentrationretrievedbyBracheretal.(2009)withthe
IOslantcolumnsintheoceanicregionsoftheworld.Bothdataproductsareretrievedfrom
theSciamachysensor.Somesimilaritiesinthespatialpatterncanbeseen.Especially,diatom
concentrationsareenhancedintheEasternPacificaswell,similartotheIOamounts.Theincrease
ofdiatomconcentrationsSouthof-30◦latitudeisnotsoprominentintheIOresult.Future
investigationswillshowifthelinkisprovidedbycausalityornot,i.e.ifthediatomsareproducing
iodineprecursorsofiftheproductsofFig.3.16aretheresultofathirdreason.

SCIAMACHY DOAS fit (OC_CHL, 15. October –14. November 2005)

-180-120-60060120180

0.01.5Absor3.0ption strength4.5(Fi6.0t-Factor)7.5 along9.0thelight10.5path1213.5
Figure3.16:Left:Absorptionstrength(fitfactor)ofthediatomconcentrationintheworld’soceans
derivedfromSciamachymeasurementsbyBracheretal.(2009)forOctober15toNovember14,
2005.ReprintedfromBracheretal.(2009).Right:CompositeofIOdataforthesametimeof
yearbutaveragedover4years.Thelandismaskedtomatchtheleftgraphandtosetfocusonthe
regions.eanicco

okOutloInGlyoxalresultsfromSciamachy,theEasternPacificalsoshowsenhancedcolumndensitiesina
regionoverlappingwiththedetectedIOamounts.ThishasbeenreportedbyWittrock(2006)and
Vrekoussisetal.(2009)andadditionallyhintsatbiologicalproductionofatmospherictracegases
inthisregion.CHOCHOhasvarioussourcesandisforexampleproducedbyoxidationofvolatile
organiccompounds,whichareexpectedtobepresentinbiologicallyactiveareas.Fromship-based
measurements,enhancedamountsofCHOCHOandIOhavebeenobservedintheEasternPacific
(RainerVolkamer,personalcommunication).Comparisonsbetweendatafromtheseindependent
studieswillbefurtherinvestigatedinthenearfuture.Itisinterestingtonote,thatsimilartothe

121

3ObservationsofIOfromsatellite

IOresults,theGlyoxalamountsarenotenhancedabovetheNorthwestAfricanupwellingregion.
Inanycase,theEastPacificseemstobespecialwithrespecttoiodinecompounds.Comparisons
withGlyoxalaswellaswiththespatialdistributionsofdifferentphytoplanktonspeciesandalso
withmeasurementsoforganohalogensintheoceanwaterswillinthefuturegivefurtherinsightinto
theseinitialfindings,whichrevealaninterestingrelation.

3.5ObservationsofIOontheNorthernHemisphere
Uptoknow,resultsofIOmainlyontheSouthernHemisphereweredisplayedanddiscussed.Before
IOamountsandspatialdistributionsontheNorthernHemispherecanbeexamined,achangein
thereferenceregionneedstobeperformed.ForgloballyconsistentIOresultsitisdesirabletouse
aconstantreferenceregionthroughoutacompletestudy.Unfortunately,thisisnotpossiblefor
thecaseofIOyet.Itwasalreadymentionedinthebeginningofthischapter(Section3.1)that
theIOcolumnsshowageneraltendencytowardssmallerandevennegativevaluesontheNorthern
Hemisphere.ThisisnotduetorealIOabsorptionbutiscausedmostprobablybygeometrical
influenceonthemeasurementdata.Changesinviewinggeometry(especiallytherelativegeometry
betweensatelliteandsun)mayleadtochangesinstraylightamountorpolarisation.Thesetwo
processesshowspectralsignaturesinthemeasurementdatawhicharenotperfectlycorrectedfor.
Untilthisisdonemorepreciselythancurrentlyimplemented,somesolutionneedstobefound
ifresultsontheNorthernHemisphereshallbeconsidered.TheidentifiednegativeslopeofIO
columnstowardstheNorthisdemonstratedinFig.3.17.Thevaluescalculatedforthisplotwere
takenfromthefouryearglobalmap(cp.Fig.3.1)andwereaveragedoverthelongitudebandfrom
30◦to150◦Eastin1◦latitudesteps.Theminimaaround-20◦,+15◦and+35◦originatefrom
thedetectedinterferenceoverclearoceanregions(cp.Sec.3.1andChapter6).Apartfromthese
minima,thenegativetrendcanbeseen.ForthefollowingdataanalysisandimagesontheNorthern
Hemisphere,adifferentreferenceregionischosen.TheselectedregionintheNorthernPacific(at
180◦±10◦Eastand30◦±10◦North)isdisplayedinFig.3.18.Forfutureanalyses,thepossibilityof
correctingthedecreasingtrendtowardstheNorth-eitherbysubtractingtheeffect(astemporary
solution)orpreferablybyidentifyingandeliminatingthecause-willbeinvestigatedinorderto
results.globaltconsistenobtainInthenextfigures,IOcolumnsforregionsontheNorthernHemisphereareshown.Figure3.19

122

Figure3.17:Fromthe4yearglobalmean,
theIOvalueswereaveragedoverlongitudes
between30◦and150◦Eastandlatitudesteps
of1◦.Thetendencytowardsnegativevalues
ontheNorthernHemispherebecomesvisible.

3.5ObservationsofIOontheNorthernHemisphere

Figure3.18:Therectangularareamarked
inregionpurpleforwtheasanalysisselectedofasIOsuitableresultsonreferencethe
Hemisphere.Northern

containsfourseasonalmapscenteredattheNorthPoleandreachinguptoalatitudeof50◦North.
Ineachmap,dataoftherespectiveseasonfromfouryearseachisincluded.Duetotheusual
lightconditionsandthelimitationoftheSZA<84◦,theautumnandwintermaps(candd)con-
tainasmallerdataamount,thanspringandsummer(aandb),whichleadstoalargerspatial
scatter.Interpretationoftheamountsisthereforedifficulthere.Especiallytheautumnmapex-
hibitsenhancedvaluesofIO,whichwillneedfurtherinvestigation.Inthesummerperiod(b),the
IOamountsremainlargelybelow3×1012molec/cm2,apartforafewexceptions.AttheWestern
coastlineofNorthernGreenlandandatthecoastoftheGulfofAlaskasomenotionsofIOamounts
appearwithvaluesaround4×1012molec/cm2.
ThemostprominentandclearfeaturesofIOareapparentinthespringtimemap(a),whereseveral
coastlinesexhibitenhancedamountsofIO.TheseregionsincludethecoastsofSouthernGreenland,
theWestcoastofSpitsbergenbetween75◦and80◦North,ontheSouthcoastofNovajaZemlja(the
NorthRussianarchipelagobetweentheBarentsSeaandtheKaraSea),thecoastlineoftheGulf
ofAlaskaandsomesmallervaluesalsointheBeringSea,overIceland,andbetweentheHudson
BayandBaffinIsland.AcloserviewforSpitsbergen/SvalbardfollowsinChapter4,Sec.4.3,asthe
BremenDOASgroupmaintainsaground-basedinstrumentinNy-Ålesund.Largestvaluesatthese
coastlinesreach6×1012molec/cm2intheaveragedspringtimeperiod.Coastlinesarefamiliar
locationsfortheexistenceofincreasedalgaepopulation.Possibly,themainsourceintheseregions
isbiologicalreleaseoforganohalogensalso.
Ifthereleaseismainlyoforganicorigin,thespatialdistributionofregionsshowingenhanced
IOvaluesdependsonthedistributionofemittingspecies.Thequestioniftheiodineisreleased
ratherbymacroalgaeand/orphytoplanktonattheNortherncoastlinesisnotclarifiedyet.
InmostofthecaseswhereIOisenhancedontheNorthernHemisphere,theorientationof
theinvolvedcoastlinesseemstohaveasystematiccomponent.FurtherNorth,theaffectedcoast
linesrunfromSouthwesttoNortheast,whilefurtherSouth,therespectivecoastsareratherfacing
towardstheSouth,e.g.attheSoutherncoastofGreenland.InFig.3.20,thisisdisplayedin
additiontothreeselectedSciamachyorbitswhichpassovertherespectivecoastlines.
Thedescribedtendencyiseitheraresultofcoincidenceoritmightalsoberelatedtospecific
oceancurrentsandrelatedaccumulationofcertainalgaespecies.Athirdoptionmighthintat
anirregularityinconnectionwiththeorientationoftheSciamachyorbits.Itisnotclearwhich

123

3

(a)

(c)

ationsObserv

of

IO

from

satellite

(b)

(d)

Figure3.19:SeasonalaveragesofIOslantcolumnsontheNorthernHemisphere,usingdatafrom
fourratherhigh,subsequenthetyspringearseac(a)h.andWhilesummerthe(b)spacialmapsconscattertaininmuthechmoreautumndata(c)pandointswinandtersho(d)wsmomapsothis
resultswithlowvaluesinsummerandenhancedIOatcoastlinesinspringtime.

124

3.6ThedifferencebetweentheArcticandtheAntarcticIOobservations

influencecouldprovokehighIOamountsinthisrespect.Instrumentalinfluenceslikethememory
effectcaninprincipleinfluencethemeasurementsatcertainregions.However,thispossibility
doesnotseemreasonableinthepresentcase,asthesequenceofSciamachymeasurementswithin
onemeasurementstatedoesnotrunfromthepossiblyice-coveredlandtothedarkeroceanat
theaffectedcoastlines,butalongthecoast.Onepossibleeffectistheimprovedsensitivityof
satellitemeasurementstowardspatternsonthegroundwhichareorientedinthesamedirection
astherectangulargroundpixel.Inthiscase,therespectivefeatureiscapturedbyonesatellite
measurementandnotdistributedoverseveralpixels.Asimilarobservationhasbeenmadein
connectionwithshipemissionsofNO2,whichbecomebettervisibleiftheirdirectioniswellaligned
withtheSciamachygroundpixels(Richteretal.,2004).ThiswouldimplythattheIOamounts
attheindicatedcoastlinesarebettervisibletotheinstrumentthanatdifferentlyorientedcoast
lines.

3.6ThedifferencebetweentheArcticandtheAntarcticIO
observations

WhencomparingtheIOobservationsontheNorthernandtheSouthernHemisphereseveraldiffer-
encescanbeidentified.
•IntheAntarctic,thedetectedIOamountsarerelativelywidespread,andcoverdifferentareas
fromseaiceregions,tocoastlinesandiceshelvesandalsothecontinent.TheIOremains
enhancedoverratherlongperiodsoftimewithavarietyoftemporalandspatialvariations.
HighestamountsclosetothecontinentappearinOctobereachyear,whiletheIOoverthe
seaiceregionsreachesitsmaximumlaterinNovemberandDecember.
•TheArcticshowsnoevidenceforwidespreadIOcolumnsabovethedetectionlimitofScia-
machy.StronglyconfinedcoastalregionsatseverallocationsontheNorthernHemisphere
exhibitenhancedIOmainlyduringspringtime.IntheArcticseaiceregions,noIOisobserved
time.springinspacefromOverall,thecomparisonbetweentheArcticandAntarcticregionsstronglysupportstheconceptof
organicreleaseofiodinecompounds.
ForBrO,wheremaximaarefoundonbothHemispheresinPolarSpring,mainlyinorganicsources
arepresumed.Thesuggestedinorganicreleaseprocessescantakeplaceinthephysicalsituations
intheArcticandtheAntarctic.Inorganicpathwaysareinprinciplealsopossibleforiodinerelease,
buttheywillbedifferentfromthebrominemechanismthen,becausethetemporalandspatial
patternsaresodistinct.SuchpathwayswillnecessarilybeconnectedtosomeAntarcticspecific
property,e.g.ofthe(sea)icestructure,butarenotknownofyet.
Organicreleaseofiodinecompoundsisknownfromalgae,phytoplanktonandalsobacteria,and
astheiodinereleaseisspeciesspecific,thedistinctbiospheresofNorthandSouthmaycausethe
differenceinIOdistributions.Somedifferenticepropertiesmightallowreleaseofbiogenicmolecules
moreeasilyintheAntarctic.IntheAntarctic,e.g.,polynyasaretypicallylargeandalsoextend
towardstheopenocean,whileintheArctictheyaregenerallysmallerandmorealongthecoasts

125

3ObservationsofIOfromsatellite

Figure3.20:ThreeSciamachyor-
bitsfromApril10thand13th(see
headerfororbitnumbers)forwhichthe
groundpixelswithinthefieldofview
ofthenadirmeasurementsaremarked
ingreen.Somecoastlineswherehigh
IOvaluesweredetectedareencircled
inredanditbecomesevidentthat
therelativeorientationofthecoast
linestotheorientationoftheScia-
machygroundpixelsisapproximately
parallel.

(ThomasandDiekmann,2003).Thismayfacilitateclosercontactoficealgaeorphytoplankton
speciestotheatmosphereaboveandhence,thereleaseofgaseoussubstanceintothepolarboundary
layerismoreprobableintheAntarctic.

3.7NoteontherelevanceoftheretrievedIOamounts

Inadditiontothediscussionsofprobablesourcesasconsequenceofthespatialandtemporalpatterns
ofIO,thesatelliteobservationsofIOmayalsobeusedtoestimatethepotentialimpactofiodine
chemistryonozonelossandparticleformation.Thesequestionshavebeenaddressedbyground-
basedmeasurements(Dickersonetal.,1999;O’Dowdetal.,2002b;McFiggansetal.,2004;Read
etal.,2008)andmodellingstudies(Vogtetal.,1999;McFiggansetal.,2000),andmostlyconclude,
thatiodinehasasubstantialimpactinbothaspectsatleastlocally.Followingfromthesatellite
observations,informationonalargerspatialscalebecomesavailable.Ifthesatelliteobservations
areusedinthisdiscussionitneedstobekeptinmindthough,thattheretrievalsyieldcolumn
densitiesandassumptionsareneededtodeterminelocalconcentrations.Theseassumptionsare
accompaniedbysometypicaluncertainties,butanestimationonanorderofmagnitudebasisis
possibleandgivessomeinsightintothepotentialimportancenevertheless.Asanexample,the
particleformationfromiodineoxidesiscomparedtothatfromsulphurprecursorsinorderto
estimatetherelativeimportanceofiodineoxides.Forthecalculations,ratecoefficientsaretaken
(2006b).al.etSanderfromHigheriodineoxidesandiodineoxideclustersactascondensationnucleiforaerosolformation
(Sec.1.5.3),whichisinitialisedbytheIO+IOselfreaction(R34andR35).Therateofhigheroxide
productionthereforedependsonthesquareoftheIOconcentration,makingthisisahighlynon-
linearprocess.Therateofiodineparticleformationr(Ipart)isdeterminedbytheformationrateof
higheriodineoxides.Astheratioofthereactions(R34)and(R35)withrespecttothetotalrateof
selfreactionkIO+IOiscloseto1(Blossetal.,2001),theformationrater(Ipart)canbeapproximated
byr(Ipart)≈(kIO+IO)[IO]2.TheratecoefficientfortheselfreactioniskIO+IO=8×10−11cm3/molec/s.

126

3.7NoteontherelevanceoftheretrievedIOamounts

Forcomparison,theformationofparticleprecursorsfromsulphurspeciesisestimated.Forthis,
H2SO4isacrucialspecies.AssubsequentreactionsgeneratingH2SO4moleculesarerapid,therate
determiningstepintheH2SO4formationchainisthethreebodyreactionofOHwithSO2(OH+
SO2+M→HOSO2+M).TheformationrateofsulphateparticlesfollowingthereactionofSO2
withOHisthengivenbyr(Spart)≈k(OH+SO2+M)[OH][SO2].Todeterminer(Spart),thereaction
ratecoefficientk(OH+SO2+M)forhighpressure(largeconcentrationofthirdbodyM)ischosenas
wellasdaytimemarineboundarylayerconditionsforSO2(100ppt)andOH(1×106molec/cm3).
Usingatypicalobservedupperlimitof3×1012molec/cm2IOverticalcolumn(fromaslant
columnof3×1012molec/cm2andanAMFof1overtheocean)andprofileheightsofwellmixedIO
intheboundarylayerbetween1kmand100m,theIOconcentrationsliebetween1.2and12ppt.
Thisyieldsarangefortheiodinetosulphurparticleratioof:

r(Sr(Ipartpart))=18...1800

Thesenumbersgivearoughestimationontheratesofparticleprecursorsubstancesfromiodineand
sulphurchemistry.ThecalculationoftheactualnumbersofCCNforminginthegivenatmospheric
conditionsneedstobeperformedusingdetailedatmosphericchemistrymodelling,includingprecise
informationoncrucialsubstancessuchasDMSforthesulphatechemistryandnucleation,andmost
importantlyalsothefinalclusteringprobabilities.Thisisrequiredtoassessaccuratelytherelative
importanceofiodinechemistryasaglobalsourceofparticles.Theestimationsabove,however,
alreadyshowthattheproductionratesofdirectiodineparticleprecursorsrangeinthesameand
higherordersofmagnitudesascomparedtothoseofthesulphateaerosol.Exactnumbersarein
strongdependenceoftheindividuallychosenconditions,sothattheresultcoversseveralordersof
magnitude.

127

4ValidationandcasestudiesofsatelliteIO

Forthevalidationofsatellitemeasurements,itisgenerallyrequiredtocomparetheresultsfrom
spacewithotherobservations,suchasground-basedorairbornemeasurementsforasmanylocations
andtimeperiodsaspossible.Ifsuitableconditionsareusedforbothdatasetsandifthenthe
resultsagreereasonablywell,suchasuccessfulcomparisonlendsconfidencetotheresultsofthe
comparedobservations.Ifnosystematicerrorisinducedwhengoingtodifferenttimesorlocations,
acomparisonstudyalsosupportsthecredibilityofthesatelliteretrievalsforsceneswherenoother
measurementshaveevertakenplacebefore.
Thefollowingconsiderationisvalidforallcomparisonsbetweensatelliteandground-based
measurementsofIO.Anexactagreementbetweenthesatelliteobservationsandground-baseddata
cannotbeexpected.Thereasonforthisisthedifferentspatialcoverageofthemeasurements.While
typicalground-basedmeasurementsprobearatherconfinedregionontheorderofmetrestoafew
kilometresatmost,thesatellitegathersallatmosphericinformationfromagroundpixelofatleast
30×60km2size.Duetothehighspatialvariabilityexpectedforthesourcesofiodinespecies,
differencesbetweenthedifferentmeasurementplatformsareexpected.Nevertheless,validationwith
ground-baseddataismeaningful,ifthefocusisnotsetontheexactvaluesbutonthespatialand
temporaltrendsandontheagreementwithinacertainrange.
InthecaseofIOmeasurements,datasetsforcomparisonfromground-basedmeasurementsor
otherplatformsaresparse.SeveralstudieshavereportedIOvaluesbelowthedetectionlimitofthe
satelliteretrievalsattherespectivelocation.Forthese,zeropointvalidationmaybeperformed,
whichmeanstoprovethattheSciamachyIOretrievalyieldsanIOvaluebelowdetectionlimit
fortherespectivetimeandlocation.
Someground-basedmeasurements,however,wereconductedinregionswhereSciamachydetects
IOabovethedetectionlimit.Someindividualcomparisonshavebeenperformedwithselecteddata
setsofsuchindependentstudies.

4.1Comparisonwithlong-pathDOASmeasurementsatHalley,
rcticaAnta

FortheAntarctic,wheresatelliteobservationsrevealfairlylargevaluesofIO,ground-basedmea-
surementshavebeenconductedduringtheCHABLIS(ChemistryoftheAntarcticBoundaryLayer
andInterfacewithSnow)campaign(Saiz-Lopezetal.,2007b,andreferencestherein)closetothe
aforementionedHalleyResearchStationfromJanuary2004untilFebruary2005.IOvolumemixing
ratiosweremeasuredclosetotheground(ataheightof4-5mabovetheicesurface)bytheactive
long-pathDOAStechniqueusingaxenonlampasanartificiallightsource.Thelightbeamtravels

129

4ValidationandcasestudiesofsatelliteIO

horizontallyoveracertaindistanceandisreflectedbackbycorner-cubereflectorstoatelescope
andspectrometerunit.TheVMRisdeterminedasaverageoverthetotalopticalpathlengthofthe
lightbeam,whichhasbeen8kmintheconsideredcase.
Foracomparisonwiththeground-baseddata,thesatelliteIOresultsaroundHalleyStation
havebeenextracted.ThefouryeartimeseriesforthislocationhasbeenpresentedinSec.3.2.2,
Fig.3.5.AllIOdatafallingintoaboxof500kmsidelengthwithHalleyStationinthecenter
havebeenusedforthetimeseries.Theareaofthesatelliteobservationsismuchlarger,butcovers
thatoftheground-basedmeasurements.SpatialvariationsinIOamountswillthereforereducethe
correlationbetweenthetwodatasets,butusingamuchsmallerregionforthesatellitedatawould
reducethedataamountwithaccordingnegativeinfluenceonthesignal-to-noiseratio.
InordertocomparetheVMRvaluesoftheLP-DOASresultswiththeslantcolumnsofthe
satelliteobservations,theairmassfactoraccordingtothesatellitegeometryandtheIOprofile
isneededforthecalculationoftheverticalcolumnandthegroundVMR.TheIOprofileisnot
knownfortherespectivelocationandtimeperiod,sothatanAMFforthesatellitemeasurements
cannotbecalculatedexactly.NoIOprofilehasbeenmeasureduptonowandonlysomegeneral
estimatesareavailable.However,withcertainassumptions,meaningfulcomparisonisnevertheless
possible.Mostimportantly,assumingthattheverticaldistributionofIOdoesnotchangeoverthe
timeperiodofthecomparison,theconversionbetweenthecolumnamountandtheVMRisgivenby
aconstantfactor.Inthiscase,theevolutionofthetwomeasurementtimeseriescanbecompared.
InFig.4.1,thetimeseriesforbothinstrumentsfromFebruary2004untilFebruary2005are
shown.ThetoppanelcontainsanexcerptoftheSciamachytimeseries(cp.Sec.3.2.2),with
dailyslantcolumnvalues(blackdots)andadditionallyaweeklyrunningmean(blacksolidline).
ThebottompaneldisplaystheIOVMRfromtheLP-DOASmeasurements,aspublishedinSaiz-
Lopezetal.(2007b).ConcurrentBrOmeasurementsarealsoshown,butnotfurtherusedhere.In
bothIOdatasets,asubstantialday-to-dayvariabilityisvisible.Thisscatterhasseveralsources.
Noiseinthemeasurementdataalreadyleadstosomespreadintheresults.Inaddition,however,
theIOamountstronglydependsoncurrentconditionsandactualemissionratesandthereforeis
indeedvariableitself.Insometimespanswithintheshownperiodadirectcomparisonisnot
possible.Duringwinter,e.g.,nosatellitemeasurementsclosetothepolearerecordedduetoalack
ofsunlight.FortheactiveLP-DOASmeasurements,thisisnotaproblem,butforothercampaign
relatedreasons,theLP-DOASmeasurementsofIOwhereonlyconductedoncertaindays.
Theevolutionofthetwotimeseriesisverysimilarandsomecharacteristicscanbeidentified.
Inbothdatasets,thelargestlonglastingamountsofIOareobservedinspringtime,especiallyin
October.Goingfromwinterrightinthemiddleofthetimeseriestowardsthisspringperiod,theIO
valuesareincreasingandafterthemaximuminOctober,somewhatreducedamountsareretrieved.
Thisfeaturecanbeidentifiedinspiteoftheaforementionedscatterinbothdatasets.Intheground-
baseddata,thevaluestrulyrangearoundzeroinwinter(e.g.inAugust)andgouptoaround7ppt
duringOctoberwithsinglespikesinthemeasurementsupto20ppt.SuchhighIOamountsare
rarelyobservedandprobablyareshort-termevents.Towardssummer,IOVMRamountsdropto
around2pptandremaininthisrangeforthewholeseason.Thetimeseriesofthesatelliteresults
beginsinSeptemberandrisesfrom3×1012molec/cm2toabout6×1012molec/cm2inmid-October

130

4.1

withComparison

long-path

ASDO

tastmeasuremen

nA,Halleytarctica

IOFiguremeasuremen4.1:tsComparisonfromtheoftheCHABLIStimeseriesofcampaignSciama(bchyottom);bobservottomationsfigure(top)reprowithducedlong-pathfromDOSaiz-AS
Lopdenoteezettheal.beginning(2007b).ofBoththemondatathsetsinbareothshocases.wnforForthethesameSciamatimepchyeriod,results,theticdailyksaonvtheeragedaxesIO
vdataaluesisliesshobewntwaseen1singleandp5oin×ts,1012awmoleceekly/cmmean2,arespssectivolideline.errorThebarsarestandardomitteddeviationforbetterofthevisibilitdailyy
ofthetemporalevolutionhere.Thelong-pathDOASshowssinglemeasurementsaspointsanda
10-daymovingaverageassolidline.

131

4ValidationandcasestudiesofsatelliteIO

withvaluesupto10×1012molec/cm2onindividualdays.Consistentwiththeground-baseddata,
thevaluesarelowerinsummerandremainpositive.Bothinstrumentsshowagainhighervaluesin
autumn(i.e.aroundFebruary-March),whichismorepronouncedinthesatelliteresults.Thedipin
midMarchisratherunusual,doesnotoccurinthelateryears(Fig.3.5),andisanindicatorofthe
strongIOvariability.ThesatellitestilldetectsquitelargeamountsinApril(somesinglehighvalues
causeajumpintheendofApril),wherenoground-baseddataareavailableforcomparison(no
datapointsbetweenbeginningofMarchandbeginningofMay).Inotheryears(cp.Fig.3.3),the
satelliteIOtrendalsoshowsdecreasingvaluesalreadyfromMarchtoApriltowardspolarwinter.
Inconclusion,theevolutionofthetwoco-locatedtimeseriesagreeswithsomeexpecteddifferences
details.theinForaquantitativecomparisonbetweenthetwodatasets,nowareasonablesurfacevolume
mixingratioiscalculatedfromthesatelliteslantcolumns.Inspiteofthemissingknowledgeabout
theexactIOprofile,confinementofIOtothelowestlayersisexpected(Saiz-Lopezetal.,2007c)and
anexampleprofileshapemaybeassumed.TheusualboundarylayerheightinAntarcticaislowas
comparedtothemarineboundarylayer.AconstantIOVMRinthelowest100mdroppingtozero
aboveassuggestedbySaiz-Lopezetal.(2007c)isareasonablechoice.Themaximumslantcolumn
amountsinAntarcticspring(aroundmidOctober)arearound6×1012molec/cm2fortheaveraged
Sciamachyobservations(Fig.4.1).ConsideringthetypicalAMFof4forAntarcticconditions(cp.
Fig.2.11),theverticalcolumnamountsto1.5×1012molec/cm2.Withtheabovechosenprofile
shape,thesevaluescorrespondstoasurfaceVMRofapproximately6ppt.Theaveragevalueof
theground-baseddatasetreaches7pptaroundmidOctober.Althoughthistypeofcomparison
essentiallyaddressestheorderofmagnitudeoftheresultsduetotherequiredestimations,thede-
viationbetweenthetwoVMRamountsisonly15%,demonstratingagoodagreementbetweenthe
results.satelliteandground-based

Overall,thecomparisonofthetwoindividualstudies,thesatelliteandground-basedobservations
atHalleyStationontheAntarcticcoast,areingoodagreementregardingthetemporalevolution
oftheIOamountsandalsoregardingtheorderofmagnitudefortheabsoluteVMRvalues.This
conclusionispromisingandstrengthensthecredibilityofthesatelliteIOretrieval.

4.2ComparisonwithanindependentstudyusingSCIAMACHY
data

SciamachynadirmeasurementshavebeenevaluatedfortheabsorptionstructuresofIOinone
independentstudy,publishedbySaiz-Lopezetal.(2007a).Fourselecteddaysofdatahavebeen
analysedandtheIOverticalcolumndensitiesfortheSouthernHemispherearereported.Inagree-
mentwiththepresentstudyhere,Saiz-Lopezetal.findenhancedvaluesofIOinAustralspringtime
aroundtheAntarctic.Apartfromthisgeneralfinding,mostofthedetailsshowlargedifferences
tothepresentworkthough.TheIOverticalcolumnsforthefouranalyseddaysarereprintedin
Fig.4.2,asoriginallypublished(Saiz-Lopezetal.,2007a).ThespatialmaximumofIOliesaround
60◦Southmostlyoverregionswhicharecoveredbyseaiceduringthattime.

132

4.2ComparisonwithanindependentstudyusingSCIAMACHYdata

Inshort,theretrievalofthealternativestudyusedafittingwindowof426-440nm.The
tracegasesNO2,O3,O4,H2O(g),andtheRingeffectaswellastheVRSinwaterwereincluded
inthefittingroutine.Inasecondretrievalrun,themeanresidualfromthefirstrunhasbeen
additionallyincludedas”pseudo-absorber”inthefit.Thespectralfittingresultsareshownonly
afterthecorrectionforthismeanresidual.Thefitqualitypriortothecorrectionisthereforenot
known.Saiz-Lopezetal.statethattheIOresultsdifferonlyby1%betweenthetworetrievals(with
andwithoutincludingtheaveragedresidual).Thisisactuallynotsurprisingastheresidualisthe
remainingpartofthemeasurementwhichdoesnotexhibitanycorrelationwiththerespectivetrace
gasesanymore.Ineachfitofanindividualspectrum,theresidualspectrumislinearlyindependent
oftheincludedabsorptionstructures(notconsideringnoise).Consequently,thisalsoholdstrue
forthesum(average)ofallresidualsoveroneorbitandtheinclusionshouldnotaffecttheresults
oftheothertracegases,butreducesthefinalresidual.Consequently,suchacorrectionprocedure
needstobeusedwithcareaspotentialretrievalerrorsmaybehidden.Therefore,thishasnotbeen
appliedinthepresentwork.

ComparingtheresultsbySaiz-Lopezetal.withtheresultsinthepresentstudyinmoredetail,
somepointsofdiscrepanciesareidentified.
Firstofall,thedeterminedIOcolumnamountsdonotagree.Thehighestwidespreadamounts
foundbySaiz-Lopezetal.rangearoundaverticalcolumnof20×1012molec/cm2(cp.redcolour
codeinFig.4.2).AsageometricalAMFisused(i.e.,AMF=1+1/cos(SZA)forexactnadirview),
alightpathenhancementofafactorof3foraSZAof60◦andexactverticalnadirviewingdirection
isconsidered.Thehighvaluesthereforecorrespondtoslantcolumnsofatleast60×1012molec/cm2,
individualpixelsexhibitslantcolumnsupto80×1012molec/cm2.Thepresentstudydetectsno
widespreadslantcolumnsover10×1012molec/cm2andonlyscatteredvaluesreachingashighas
20×1012molec/cm2.
Secondly,thespatialpatterndiffers,whichbecomesapparentwhenconsideringthemonthly
meansforthestandardIOproductofslantcolumnswhichhavebeendiscussedabove.InSaiz-Lopez
etal.(2007a)theenhancedvaluesofIOareconcentratedatsomedistancefromtheAntarcticcon-
tinent,whereatthattimeofyeartheoceaniscoveredwithseaice,andalsobeyondtheseaicein
patchesovertheopenocean.ThefouryearaveragedmonthlymeansaroundOctoberinthepresent
studyarealsoenhancedoverseaice,butclosertotheAntarcticcoast,andadditionallyoverthe
iceshelvesandsomepartofthecontinent.EnhancedIOvaluesovertheseaiceatfurtherdistance
fromthecoastappearaboutonemonthlater(cp.Fig.3.7).

Foronesampleday(October5th2005),someviewsontheSouthernHemispherearecombined
inFig.4.3whicharehelpfulforthefollowinganalysis.Field(a)inthisfigureshowsthestandardIO
product(V1.28)forthatday.Forthepurposeofbettercomparability,thesameroughgeometrical
AMF(neglectingtheinfluencesofatmosphericscatteringandsurfacereflectance)isappliedand
verticalcolumnsareshown.Apparently,thedailyvaluesexhibitlargediscrepanciesbetweenthe
twostudies,asnoenhancedverticalcolumnscanbedistinguishedforthestandardIOfitinthis
colourscale.ThecolourscaleinFig.4.3(a)ischosentobethesameasinFig.4.2fortheresults

133

4

alidationV

and

case

studies

of

satellite

IO

Figure4.2:FigurereprintedfromapublicationbySaiz-Lopezetal.
SouthernHemisphere(upto30◦South)showsonedayofSciamachy
forfourconsecutivedays(4th-7thOctober,2005).

134

results(2007a).ofEacIOhslanmaptofcolumnsthe

4.2ComparisonwithanindependentstudyusingSCIAMACHYdata

fromSaiz-Lopezetal.(2007a).Fordirectcomparison,Fig.4.3(b)repeatsthemapforOctober5th
.studythatfromThebasicdatasetofsatellitespectraislargelythesameinbothstudies,althoughitisnot
known,whichcalibrationsteps(cp.Sec.2.1)weresetandwhichwereomittedbySaiz-Lopez
etal.(2007a).Thequestionarises,whichdifferencesinthedataanalysisproceduremaylead
totheobserveddifferencesinIOresults.AsdiscussedinSec.2.8,thechosenwavelengthrange
fortheanalysismaybeasourceofdifferences.TheretrievalofSaiz-Lopezetal.cannotbe
exactlyreproduced,assomesettingsintheretrievalalgorithmarenotrevealed.However,from
themultitudeofdifferentretrievalrunstestedinthepresentstudy,someparameterversionshave
similarsettingsandshowsimilarresultsascomparedtoSaiz-Lopezetal..Oneexampleoftheseis
discussedinSec.2.8asVersionV0.27iandusesthespectralwindowfrom418-438nm.Thefitting
windowofSaiz-Lopezetal.alsoincludesthe430nmregion,whereforV0.27iretrievaldifficulties
tified.idenerewFigure4.3containsresultsfromtheV0.27iretrievalinfields(c)and(d).Comparisonwith
resultsfromSaiz-Lopezetal.(2007a)highlightssomepointsofconcernassomesimilaritiesare
revealed.ThevaluesoftheV0.27iretrievalyieldhigherIOcolumnsthanthestandardIOfit.
WidespreadamountsofIOreachuptoabout10×1012molec/cm2verticalcolumndensity,visible
inFig.4.3(c).Theseamountsarebyafactorof2smaller,butstillcomparabletotheamounts
observedbySaiz-Lopezetal..ThecolourscaleinFig.4.3(c)isalsoadaptedtotheoneusedin(b).
Inmostlocations,thepatternoftheV0.27ioutputagreesremarkablywellwiththeresultsin
field(b).ThewidebandofenhancedIOvaluesiswellreproducedintheV0.27ifithere.Except
forveryfewpixels,mostofthespatialstructuresofhighIOintheindividualstatesofsatellite
measurementscanbeseeninbothdatasets.Ingeneral,suchafindingwouldgiveanindication
thatthetwodatasetscapturetherealitywell,butfollowingfromthediscussioninSec.2.8ithas
beenconcluded,thattheV0.27ifitexhibitsconsiderableretrievalerrors.Possibly,thisholdsalso
truefortheexternalstudy.Figure4.3(d)containstheresidualrmsvaluesoftheV0.27iretrieval.
Comparisonoffield(d)with(c)and(b)demonstrates,howsimilarthespatialpatternsofhighrms
(i.e.poorfitquality)fromretrievalV0.27iareincomparisontotheenhancedIOvaluesseenby
thisretrievalaswellasbySaiz-Lopezetal..Comparableinputinformationisusedinbothstudies,
sothatsimilarproblemsmayoccur.
Oneadditionalcuriousfindingisidentifiedonthe5thofOctober.Onthisday,enhancedIO
valuesareobservedfaroutsidetheAntarctic,SouthwestofSouthAmericainthestateofsatellite
databetween70◦and90◦Westand50◦and60◦South(cp.Fig.4.3bandc).Thecolumnshereare
justaslargeasthoseovertheseaice,butthisregionisicefree.Forthecaseofanunderlyingocean,
thetrueverticalcolumnsherewillbeevenhigherthanthoseobservedoverseaiceduetoasmaller
AMFoverdarksurfaces(cp.Sec.2.3).However,abrightcloudcoverwasidentifiedforthisscene
exhibitingthesamepatternastheenhancedIOvalues.FortheV0.27iretrieval,thecorrelationof
highIOamountsandtheoccurrenceofcloudsorsimilarcaseswithhighlyreflectingsurfaceshas
beendetected(cp.Sec.2.8).
Theconclusionisambivalent.Inanycase,theactualqualityoftheexternaldatasetcannot
befullyevaluatedandjudgedhereduetoalackofdataamountandinsightintotheretrievalper-

135

(a)

4ValidationandcasestudiesofsatelliteIO

(c)

V1.28 IO: 05.10.2005

V0.27i IO: 05.10.2005

(b)Result from Saiz-Lopez et al. (2007a)VC IO

d)((d)

Figure4.3:CombinationoffourplotscomparingretrievalresultsontheSouthernHemispherefrom
differentransformedtstudiestovforerticalthedatecolumnsofbyOctobtheer,same5th,geometrical2005.(a)AMFshowsastheusedbystandardSaiz-LopIOezproetductal.(2007a),(V1.28)
forretrievwhicalhtheV0.27iretrievobtainedalresultintheisreprinpresenttedwinorkfieldand(b).discussedFieldsin(c)Sec.and2.8.(d)(c)conshotainwstheoutputIOvfromerticalthe
columnresidualinrmsthevalue.sameThecolourrmsscalerevaseals(a)highandv(b),alues,whilewhere(d)IOamoundemonstratestsarethefitenhancedqualitinyfieldsthrough(b)andthe
(c).

136

4.3Comparisonswithground-basedpassiveDOASmeasurements

formance.AmainpointofconcernisthesimilaritywiththecorruptV0.27retrievalofthepresent
studyincombinationwiththeidentificationoflargeresidualsandunrealisticresultsoverclouds
andhighlyreflectivescenes.ThisobservationseemstobereproducedinthestudybySaiz-Lopez
etal..Theirresultsmayprovetobeaccuratenevertheless.Inthatcaseitwillbenecessaryto
clarifyhowtheV0.27iresultscanleadtocomparableIOamountsinspiteofsystematicerrors.The
highvaluesovercloudswouldintroducenewquestionsontheproductionandpathwaysofiodine
ecies.sp

Resultingfromtheaboveanalysis,thediscrepanciesbetweentheindependentstudybySaiz-Lopez
etal.(2007a)andthepresentstandardIOretrievalarerespected,butshouldnotleadtoserious
doubtsintheV1.28standardfituntilmoreinformationontheindependentstudybecomesavailable.

4.3Comparisonswithground-basedpassiveDOASmeasurements
TheUniversityofBremenDOASgroupmaintainsground-basedDOASinstrumentsinseverallo-
cationsatvariouslatitudes.Additionally,someinstrumentsareoperatedoncampaignbasisat
certaintimesandlocations.OneinstrumentispermanentlyinstalledatNy-ÅlesundonSpitsber-
gen,Svalbard(78◦55’North,11◦57’East,cp.Fig.4.4),providingalong-termdatarecordsince
1995inzenithviewinggeometry.Afterseveralupgradesoftheinstrument,furtherviewingangles
wereaddedwith5viewingdirectionssince2002and11directionssince2006.

yN-Ålesund

albardvS

wingshoMap4.4:FigurethelocationoftheDOASin-
strumentatNy-Ålesundon
theislandofSpitsbergen,be-
longingtoSvalbardonthe
Hemisphere.Northern

MeasurementsfromthisstationhavebeenanalysedfortheabsorptionofIO,andonseveral
days,amountsabovethedetectionlimithavebeenobserved(Wittrocketal.,2000;Oetjen,2009).
TheinterpretationofthedatasuggeststhatsmallamountsofIOarepresentinspringandsummer.
ResultsfromtheIOobservationsinNy-ÅlesundarereprintedinFig.4.5,showingdataretrieved
byHilkeOetjen(Oetjen,2009).Inthisplot,theslantcolumnamountofIOforviewinganglesof
2◦and3◦abovethehorizonareshownwithrespecttothetimeofyear.ThepresentIOvaluesare
ratherlow,sothatoutofthe5years,onlyon29daysreliableIOamountsabovethedetectionlimit

137

4ValidationandcasestudiesofsatelliteIO

2007 2004 2008 2005 2006 reliable fit quality: all years

3 2006 reliable fit quality: all years
]-22 molecules cm1130IO dSCD [10-1JanFebMarAprMayJunJulAugSepOctNovDec

Figure4.5:IOslantcolumnsmeasuredwiththeground-basedMAX-DOAStechniqueoverseveral
◦◦ykindlyearsinprovidedNy-AlesundbyHilkate2Oetjen*(2007-2008)andFolkandard3Wittro(2004-2006)ck,UniveleversitationyofabovBremen,ethethehorizon.plotwasTheadapteddatais
fromOetjen(2009).*NowattheUniversityofLeeds,UK.

havebeenobserved.Theannualcycleapparentintheground-baseddatahasnotbeenconsidered
significant,soitwillnotbefurtherinterpretedhere.ThemainfocusliesontheretrievedIO
amountsondayswithreliablefitquality.TheslantcolumnsfromFig.4.5lieintherangeof
1-2×1013molec/cm2.
Incomparingtheground-basedresultswiththesatelliteDOASresults,themostimportantissue
andasourceoferrorsistheestimationoftheairmassfactorsforbothcases.Bothmeasurement
geometriesyieldacasespecificslantcolumnasretrievalresult.Inordertocalculatethevertical
columnandpossiblythesurfaceVMRforcomparison,asuitablecalculationoftheairmassfactor
needstobeperformed.Thecalculationissubjecttotheestimationsofparameterslikethesurface
reflectanceandthenotwellknownverticalprofileofthetracegas.Thecalculationofatypical
AMFforground-basedmeasurementsat2◦and3◦LOSabovethehorizonyields25.5and19.1,
respectively.ThesevalueshavebeencalculatedusingSciatranwithanSZAof80◦,analbedoof
90%forbrightsurfacesandasimpleRayleighatmosphere.Consequently,theverticalcolumninthe
ground-basedmeasurementsisapproximately0.5-1×1012molec/cm2.Thisvalueisnowcompared
results.satellitethetoThedetectionlimitdiscussedinSec.2.4hasanupperlimitofabout7×1012molec/cm2fora
singleSciamachymeasurement.TheevaluationoftheSciamachydatashowninSec.3.5yields
adataamountofabout280datapointsperlo√cationinthe4-yearaveragedseasonalmaps.The
reductionofthestatisticalerrorbyafactorof280,wouldindicateaminimumdetectionlimit
-ignoringallsystematiceffects-ofaround4.2×1011molec/cm2.Consequently,theobservations
fromtheground-basedmeasurementsarejustslightlyabovetheIOdetectionlimitofthelong-term

138

4.3Comparisonswithground-basedpassiveDOASmeasurements

averagedSciamachyretrievals.Thesystematicpartoftheuncertaintiesisofcoursenotreduced
bythelongaveragingperiod.
(b)(a)

(c)

(b)

(d)

Figure4.6:SeasonallyaveragedIOresultsfortheareaaboveSpitsbergen,Ny-Ålesund.InSpring
(b)(a),arehighbeloIOwamoundetectiontsapplimit.earInalongthetheotherSouthtwowestseasonssideof(c,thed),theisland,noisewhileismuthechamounstronger,tsininwinsummerter
mostoftheareaistoodarkforobservations.

ConcerningtheIOamountsretrievedfromtheSciamachymeasurementsoverNy-Ålesund,the
seasonalaveragesforfouryearshavebeeninvestigatedinSec.3.5asoverviewmapsfortheNorthern
Hemisphere(Fig.3.19).AcloseupoftheislandsofSvalbardisplottedinFig.4.6,showingthe
samefouryearaveragesasinChapter3.TheconfinementofthehighIOvaluestothecoastin
Fig.4.6(a)isvisible.ThereareonlysmallIOcolumnsinsummer(b),strongnoiseinfluencein
autumnandwinter(c,d)withhardlyanydatainwinterduetodarkness.Maximumaveragevalues
closetoNy-Ålesundliearound4×1012molec/cm2.WhiletheoceansurroundingSvalbardismostly
ice-free,thesurfacespectralreflectanceoverthesnowandicecoveredlandislarge.Dependingon
wheretheIOismainlypresentinthesatellitescene,theAMFisstronglyinfluencedbythevalues
ofthesurfacereflectance.TheAMFvariestypicallybetweenabout1.0and4.0forthesatellite

139

4ValidationandcasestudiesofsatelliteIO

measurements(cp.Sec.2.3).Foraslantcolumnof4×1012molec/cm2,theverticalcolumnthen
liesbetween1and4×1012molec/cm2.
Onlyaroughcomparisonispossiblehere,asforbothinstruments,theobservedIOamounts
areclosetothedetectionlimit.Theverticalcolumnfortheground-basedobservationsofabout0.5-
1×1012molec/cm2islowerthanthesatellitevalueof1-4×1012molec/cm2,butthetworangesmeet
at1×1012molec/cm2.IftheappropriateAMFisratheronthehighside,thevaluesagreewithinthe
statisticaluncertaintiesofthesatellitemeasurementsalone,whichare0.4×1012molec/cm2here.
Consequently,withverticalcolumnsaround1×1012molec/cm2,theIOamountsseenbythe
twoobservationsfromthegroundandfromsatellite,arecomparablysmall.Especially,typical
valuesfoundintheAntarcticaarehigher.Thisisinagreementwithfindingsatotherlocationsin
theArctic,forwhichlowamountsofIOhavebeenreportedalso(Hönningeretal.,2004).

4.4IOinmid-latitudecoastalregions

Severalground-basedstudieshaverevealedperiodicallyenhancedIOincertaincoastalcites,e.g.
atMaceHead,IrishWestcoast(Alickeetal.,1999;Carpenteretal.,2001)andinRoscoff,French
Atlanticcoast(Whalleyetal.,2007).Intheselocations,highamountsofIOintheboundary
layershowapositivecorrelationwithsolarirradiationandanegativecorrelationwithtidalheight.
Hence,thehighestamountsofIOcanbeobservedinsunlitperiodsandatlowwaters.
Thequestionaboutthesourcesoftheenhancediodineconcentrationsunderthementioned
conditionswasansweredbythediscoveryoflargefieldsofalgaealongthecoast.Certaintypesof
algaewerefoundtoproduceseveraldifferentorganohalogens,whicharereleasedtotheatmosphere
ifthealgaeareexposedtooxidativestress.Thisarisesatlowtidewhenthealgaefieldisnolonger
watercoveredbutgetsindirectcontacttotheboundarylayerair.Halogenatedorganiccompounds
arethenemittedaspartofadefensemechanismoftheplanttoprotecttissuefromoxidative
damage.Subsequently,thehalocarbonsarephotolysedandatomichalogensentertheatmosphere.
Afterreactionwithozone,speciessuchasIOareproducedandcanbedetectedbyspectroscopic
methodsinthevisiblewavelengthregion.
ThesatellitedetectionlimitforIOinthemid-latitudecoastalregionsliesafewtimeshigherthan
inthepolarregionsasdiscussedinSec.2.4.Nevertheless,thequestionarises,iftheIOobservations
fromSciamachyshowasignificantlyhigherIOamountinlowtidesituationsascomparedto
hightideperiods.Forthispurpose,theSciamachyIOdatahavebeensortedaccordingtothe
tidalphases.Firstofall,anIOtimeserieshasbeenextractedforachosenlocationonEarth.
ThestandardIOproduct(VersionV1.28)isusedhere.Inasecondstep,thetidaltimesofhigh
water(HW)andlowwater(LW)havebeengeneratedasdescribedinthenextsection.EachIO
observationrecordedinacertaintimeintervalaroundthishighorlowwatertimeisassignedtoa
respectiveHWorLWseriesofIOdata.Thissortingprocedureisperformedformorethanfour
yearsofSciamachyIOobservationsandforthetwolocations.Followingfromtheground-based
observations,theLWseriesshouldexhibithigherIOamountsthantheHWseries.

140

4.4IOinmid-latitudecoastalregions

SHOMfromdataTidal4.4.1Thetimeandheightinformationaboutthehighandlowtidesatsomechosenlocationswasgener-
atedusingatoolfromSHOM,theFrenchHydrographicandOceanographicalserviceofthemarine
(ServiceHydrographiqueetOcéanographiquedelaMarine),providedontheirdatapage1.
Thetidaldatahasaprecisionofafewcentimetresfortheheightsandafewminutesforthe
timesgiven.Thetidalheightinmetersisgivenrelativetoazerolevelwhichisroughlythelevel
ofthelowtides.Predominantly,thetimesofhighandlowwatersisofinterestandnotsomuch
theexactheightofthewaterline.Severalminutesofinaccuracycanbeaccepted,asthetiming
ofenhancedIOisnotlimitedtoafewminutes.TheobservationsreportmaximaofIOwhichare
presentforperiodsontheorderofafewhours(e.g.between0.5-3hoursinthestudyofCarpenter
etal.(2001)).Intheselectionprocedureofsatellitedatafortherespectivelocationacertaintime
intervalaroundthehighandlowtideswillbeallowedinanycase.Anexampleofthedataformat
asprovidedbytheSHOMdatapageisshowninTab.4.1.Datafromtheyear2004untilspring
selected.saw20082007SEPTEMBREDatematinpleinesmerssoirmatinbassesmerssoir
sam107h31temps8.92mhauteurtemps19h508.97mhauteurtemps01h400.84mhauteurtemps13h561.12mhauteur
lundim3208h4908h088.09m8.58m21h1720h317.79m8.45m03h0002h201.92m1.28m15h2014h372.26m1.60m
mermar5410h5109h396.98m7.51m23h5722h206.67m7.11m04h4603h453.39m2.68m17h3216h143.52m2.97m
ofTablethe4.1:marineExample(SHOM),oftheherefortidalthedataloprocationvidedofbyRoscofftheontheHydrographicFrenchandNorthwestOceanographicalcoast,listingservicethe
times(temps)ofhightides(pleinesmers)andlowtides(bassesmers)inthemorning(matin)and
theafternoon/evening(soir)of1-5September,2007.

4.4.2CasestudyforlocationMaceHead
MaceHeadisanatmosphericresearchstationontheIrishcoast(53.3◦N,9.9◦W)whereseveral
studiesonatmosphericiodinehavebeenconducted(Alickeetal.,1999;Carpenteretal.,1999;Saiz-
Lopezetal.,2006).Here,ananti-correlationofIOwithtidalheighthasbeenfoundbyCarpenter
etal.(2001)whichapproximatelyfollowsanexponentialrelation:[IO]=2.6×exp(-[TH]/1.7),with
thetidalheightTHgiveninmetersandtheIOmixingratioinppt.Forthetypicaltidalvariation
between0and5m,theIOmixingratiovariesbetweenover2pptandcloseto0ppt.
Inthesatellitedata,notonlytheexactpointintimeofhighandlowtidesisconsidered,sothe
differencebetweenthetwocaseswillbesomewhatsmaller.Stilladifferenceofaround1-1.5pptis
ected.exp2pptofIOwouldyieldatypicalslantcolumnof5×1012molec/cm2incaseofa1kmboxprofile,or
5×1011molec/cm2fora100mboxprofile(cp.tocalculationspresentedinSec.2.3andSec.2.4).As
1tmttp://www.shom.fr/fr_page/fr_serv_prediction/ann_marees.hh

141

4ValidationandcasestudiesofsatelliteIO

listedinTab.2.4,thedetectionlimitofSciamachyliesbetween2.8pptand35pptdependingon
thealtitudeprofile.OnlyincasetheIOispresentoveraconsiderablealtitudeandspatialground
range,theenhancementduringlowtidesisexpectedtobecomevisibleinthesatellitedata.
FromtheIOstandardproductofthesatellitedata,firstofall,anIOtimeserieshasbeenex-
tractedaroundtherespectiveresearchlocationandallmeasurementstakenwithinaboxof100km
sidelengthareused.ThetimeseriesusedinthefollowingarebasedontheSciamachymea-
surementsbetweenJanuary2004andFebruary2008.Insubsequentsteps,thetimeserieshave
beensubjectedtoselectionandsortingcriteriainordertoinvestigatedifferencesbetweenhigh
andlowtidesituations.Eachselectioncriterionreducesthedataamountbyacertainfactor.An
approximatereductionfactorisgivenforeachofthecriteria:

•Onlythoseobservationswithinacertaintimeintervalaroundthepointsofhighandlowtide
areretained.Fortheintervaleither±1houror±30minuteswaschosen.Thereduction
factorsfortheindividualseriesare6and12,respectively.

•Eitherthetimeseriesoftheentireyearisretainedoronlydatafromthesummerhalf-
year(April-September),whereground-basedobservationsreportedthelargestvalues.The
reductionfactoriseither1or2.

•Theinfluenceofcloudsmightbeimportanthere,soacloudscreeningwasimplementedand
couldbeswitchedonoroff.Thecloudscreeningisimplementedbyanintensitycriterion,
highreflectivesituationsareexcluded.Thisleadsroughlytoareductionfactorof2.

Dependingonthecombination,thenumberofdatapointsintheoriginaltimeseriesofScia-
machydataisconsiderablyreduced.ThisispartofthediscussioninSec.4.4.4.
TheIOdatasortedaccordingtoabovecriteriaarefinallyaveragedtoonerepresentativevalueso
thattheresultsofthehighandlowtidecasescanbecompared.Theresultsaresummarisedin
Tab.4.2,statingtheselectioncriteriaandtheresultingIOaverageslantcolumntogetherwithits
deviation.standardOneimportantobservationisthatallresultingIOvaluesarerathersmall.Takingthecomputed
standarddeviationintoaccount,alloftheIOslantcolumnsareinagreementwithzero.Regarding
theaboveestimationofexpectedIOamountsincomparisonwiththesatellitedetectionlimit,this
resultisnotsurprising,astheexpectedamountsareclosetoorbelowthedetectionlimit.Comparing
theindividualvaluesoftheHWandLWresults,bothsituationsoccur,caseswheretheHWvalue
islargerandotherswithhigherLWvalue,whiletheground-basedobservationsfindclearlyhigher
IOamountsfortheLWthanfortheHWsituations.Thetwoextremetidalsituationsareinequal
withintheirstandarddeviationsinalltestedcases.Consequently,nodifferenceinIOamountsfor
highandlowtidescanbeobservedfromthecurrentSciamachyobservations.Furtherpointsof
discussionsfollowbelow.Before,asecondexampleofcoastallocationsisinvestigated.

4.4.3CasestudyforlocationRoscoff
RoscoffissituatedattheAtlanticcoastinNorthwestFrance(48.73◦N,3.98◦W).Insummer2006,
Whalleyetal.(2007)measuredIObyLaserInducedFluorescence(LIF)inRoscoff,duringthe

142

4.4IOinmid-latitudecoastalregions

SeasonsincludedTimeintervalCloudscreeningIOathightideIOatlowtide
all±1hourno1.55±4.441.63±4.37
all±0.5hourno0.72±4.371.36±4.90
summer±1hourno2.51±3.231.53±4.17
summer±0.5hoursno1.27±2.001.55±4.31
all±1houryes1.25±5.911.21±4.95
all±0.5hoursyes-0.63±6.600.85±5.88
summer±1houryes3.07±3.470.81±4.01
summer±0.5hoursyes1.55±2.040.76±4.37

Table4.2:ComparisonofaveragedIOslantcolumns(inunitsof1012molec/cm2)forhighand
lowtideatMaceHead,Ireland,basedonsatellitedatarecordedbetweenJan2004andFeb2008.
Differentconditionsconcerningtheincludedseasons(fullyearoronlysummerhalf-year),thetime
intervalaroundthepointofhighandlowwaters(onehourortwohours)aswellascloudscreening
(viaanintensitylimit)havebeeninvestigated.Theuncertaintyrangesstatethestandarddeviation
ofallinvolvedIOdatapointsandshowthatthevaluesanddifferencesarenotsignificant.

projectReactiveHalogensintheMarineBoundaryLayer(RHaMBLe),atmaximummixingratios
around30pptforshortintegrationtimesof10s.Longerintegrationtimesyieldmuchsmaller
IOamountsdemonstratingthehighvariabilityofIOconcentrationsandthetransitorynatureof
ts.amounIOhighoutstandinglyThesatelliteIOdataisprocessedinexactlythesamewaywithidenticalselectioncriteriaas
chosenforthecasestudyatMaceHead.Ananalogousoverviewliststheresultsoftheselectionand
sortingproceduresinTab.4.3.Likeabove,alsoforthelocationofRoscoff,allIOcaseaveragesare
inagreementwithzeroIOloadtakingintoaccounttherespectivestandarddeviations.Incontrast
totheMaceHeaddata,theIOamountsforlowtideareinallbutonecombinationhigherthanthe
HWdata.Althoughthisisingeneralagreementwiththeground-basedobservations,thisfinding
isnotmeaningful,becausethisdifferencebetweenthehighandthelowtideIOamountsisnot
significant.Allamountsagreewithintheirstandarddeviations.

SeasonsincludedTimeintervalCloudscreeningIOathightideIOatlowtide
all±1hourno1.76±5.211.89±4.36
all±0.5hourno1.69±5.902.67±4.29
summer±1hourno-0.38±5.371.87±4.62
summer±0.5hoursno0.27±7.151.74±4.32
all±1houryes2.25±6.761.74±4.59
all±0.5hoursyes2.06±7.502.83±4.98
summer±1houryes0.43±7.271.32±4.43
summer±0.5hoursyes1.07±8.921.83±5.03

Tableaccording4.3:toTab.Comparison4.2,IOofslanavteragedcolumnsIOareamoungivtseninfor10high12andmolec/cmlow2tide.atRoscoff.Allsettingsare

143

4ValidationandcasestudiesofsatelliteIO

4.4.4Discussionofthetidalanalysis
Inthemid-latitudinalcasespresentedabove,thesatelliteIOcolumnshavebeenfoundtoagreewitha
zerocolumnamountwithintherespectivestandarddeviationsoftheextractedtimeseries.Reported
ground-basedmixingratiosaremostlybelowthedetectionlimitofthesatelliteobservations,so
thattheaboveresultscanberegardedasconfirmationoftwoaspects.Firstofall,thesatellite
observationsconfirmthatnolargeandwidespreadIOamountsabovethedetectionlimitofthe
satelliteinstrumentarepresentattheinvestigatedcoastallocations.Secondly,thefinding,thatthe
detectionlimitisnotsignificantlyexceeded,arguesforaconfinementoftheIOmoleculeseitherto
thelowestatmosphericlayers,orforstronglocalisation.IfIOvaluesintherangeofsurfacelevels
wereconstantlymixedoveralargealtituderange(>1km),thesatellitedatawouldcontainsome
evidenceofthis.Additionally,theIOamountsobservedbyground-basedpointmeasurementsare
effectivelyreducedbythesizeofasatellitepixelanddetectionofIOismostimpededbyprobably
sources.IOconfinedAlthoughalongdatasethasbeenconsidered,thecombinationofnecessaryselectioncriteria
stronglyreducesthenumberofdatapoints.Theindividualreductionfactorshavebeenstated
aboveandfollowingfromthis,oneselecteddataseries(eitherHWorLW)containsanumberof
datapointsreducedbyafactorof6uptoafactor48.Thelargestoriginaltimeseriescontains
nearly500datapoints,soaftertheselection,ontheorderofbetween90(weakselectioncriteria)
and10(strictcriteria)remain.ThedatabasisforanalysisisthusnotlargeandastheIOamount
isclosetothecalculateddetectionlimit,thisdataamountisnotsufficienttoreducetheinfluence
ofnoiseerrorsstronglyenough.Consequently,innoneoftheinvestigatedsituations,thedifference
betweenthehighandlowtideissignificant.
Concludingfromtheabovefindings,adifferentanalysisstrategyisneededfortheobservationof
atidalsignalintheatmosphericIOloadingandfortheidentificationofsourcesaswellastheirspatial
distributionandsourcestrengths.Higherspatialresolutionandmorefrequentmeasurementswould
befavourablefortheanalysisofthesequestions.ThesatelliteinstrumentOMI(OzoneMonitoring
Instrument),animagingspectrometerinstalledontheEos-Aurasatellite(Leveltetal.,2006),
observesinsomewhathigherspatialresolutionwithatypicalground-pixelsizeof13×24km2.
However,uptonowtheobservationofIOhasnotbeenreportedfromthatinstrument.Fora
purposefulinvestigationoftheabovetasks,specificaircraftmeasurementsareproposedandplanned
forthenearfuture.Aircraftobservationsformanintermediatebetweenground-basedandsatellite
platforms.

144

5ModelingofatmosphericIOwiththe
decoCAABA/MECCA

Inadditiontoatmosphericmeasurementsfromdifferentplatformsitisofinteresttoapplyatmo-
sphericchemistrymodelstocalculateconcentrationsandvariationsofatmosphericcomponentsfor
certainconditions.Inthefieldofiodineresearch,stillmanyopenquestionsremain,especiallyon
thesourcesofreactivegaseousiodinecompoundsandonsomereactionpathwaysandtheirkinetics.
Forexample,thesourcesandpathwaysleadingtotheobservedAntarcticIOobservationsarenot
wellknown,althoughideasforexplanationscertainlyarebeingdeveloped(cp.Sec.3.3).Inaddi-
tiontotheseideas,itisilluminatingtousemodelstudiestoinspectthecurrentknowledgewhich
isincorporatedinthemodelbyanalysingwhetherornotcertainobservationscanbereproduced.
Potentialshortcomingsinthecurrentunderstandingmaybeidentifiedinthisway.
Inthepresentstudy,theCAABA/MECCAmodelcodedevelopedattheMaxPlanckInstitutein
Mainz(Sanderetal.,2005)isapplied.Amodelisneededthatincorporatestroposphericchem-
istryincludinghalogenchemistryinthemarineboundarylayeraswellasinPolarRegions.For
thecorrectcomputationofiodinecompounds,apartfromgasphasechemistry,themodelshould
includethemostimportantreactionsintheaqueousphaseandheterogeneouschemistry.Iodine
anditscompoundsshowarelativelylargevarietyofreactionpathwayssothatalargesystem
ofreactionkineticsneedstobetakenintoaccount.Therequiredpropertiesareprovidedbythe
del.moCAABA/MECCAInthefollowingsections,themodelwillbeintroducedanditsmostimportantpropertiesandchar-
acteristicswillbedescribed.Afteroutliningtheobjectiveofthismodelingstudy,thespecificchoice
ofparametersandmodelsettingsismotivated.ThemodelresultsfortheIOconcentrationsare
presented,andmodelfindingsarecomparedwiththesatelliteresults.Fromthis,conclusionson
thenecessaryamountsofemittediodineprecursoramountsareobtained.

5.1Modellingstudiesofiodinechemistryintheliterature

Severalstudieshaveassessedthetopicofatmosphericiodinechemistrywithdifferentspecificques-
tionsandincludingdifferentmodelsettings.Dependingontheunderlyingquestionsasked,the
concentrationsofsomesubstancesarepredefinedandtheirinfluenceonotherspeciesisinvesti-
gated.Someofthepreviouslypublishedmodellingactivitiesaresummarisedhere.
TheroleofiodinephotochemistryhasfirstbeenaddressedbyChameidesandDavis(1980)for
thetroposphereandbySolomonetal.(1994)alsoforthestratosphere,withamainfocusonthe
destructionofozoneinthepresenceofiodine.ChameidesandDavis(1980)considerCH3Iassource

145

5ModelingofatmosphericIOwiththeCAABA/MECCAcode

ofIatomsandcalculatetheeffectonO3,OHandNOxlevelsatamarinelocationat30◦latitude.
Theyidentifyseveralozonedestructionpathwaysandcalculatetheabundancesofdifferentiodine
compoundstolieinthepptrangeforgivenCH3Iconcentrationsupto50ppt.Duetotheimportant
troposphericroleofOH,itssourcesandsinksneedtobeunderstoodandquantified.Depending
onprecursorlevelsbetween10and50pptofCH3I,theOHconcentrationswerefoundtochange
substantiallyby22%and100%,respectively.
InmodelcalculationsbyVogtetal.(1999),emissionsofdifferenthalocarbons(CH2I2,CH2ClI,
andC3H7I)inadditiontomethyliodideaccountforatmosphericiodinelevels.Theyconsidered
certainiodinecompounds(HI,HOI,INO2,andIONO2)asreservoirsoftemporarynature,which
actedasfinaliodinesinksinpreviousstudies(ChatfieldandCrutzen,1990;Davisetal.,1996).
Thisincreasesthedegreeofcomplexityofthechemicaliodinecycling.Oneimportantresultof
theirmodelstudiesistherealisation,thatthephotochemicalreleaseofBrandClatomsfromthe
seasaltphaseisfacilitatedinthepresenceofiodine.Inadditiontotheimpactofiodineitself,the
enhancedbromineandchlorinelevelsleadtoanacceleratedozonedestructionascomparedtothe
modelbeforeincludingiodinechemistry(SanderandCrutzen,1996).
InstudiesconductedbyMcFiggansetal.(2000),observationalconstraintsofvarioustrace
gases,alsoofIO,havebeenappliedandtheimpactontheoxidationcapacityandseveralspecific
compoundshasbeeninvestigated.Onefocusoftheirworkliesontheconnectionofiodinetothe
aerosolsphase,andtheycalculatehighenrichmentfactors(severalordersofmagnitude)ofiodine
inaerosol,inagreementwithobservations.Additionally,theimpactonozonedestructionand
denoxification(lossofNOx)isonceagainaddressed.Intheirmodelruns,theyuseaconstant
concentrationofiodocarbons.
TheeffectofiodineontheHO2/OH-ratiohasbeenstudiedinmoredetailbyBlossetal.
(2005)whofindthatthosestudiesneglectingtheIO+HO2→HOI+O2reactionoverestimatethe
HO2concentrationsascomparedtoobservations.Intheirmodelcalculation,thepresentiodine
amountsleadtosubstantiallossofHO2andsomegaininOH,sothatboundarylayerchemistryis
considerablyaffectedbyevensmallamountsofiodine.
Whilemostmodelworkisadjustedtomid-latitudeconditionsofthemarineboundarylayer,
recently,Saiz-Lopezetal.(2008)haveproposedamechanismforthereleaseofiodinespeciesin
sea-icecoveredregions,whereemissionsfromicealgaebelowtheicesheetsplayamajorroleand
multi-phasechemistryisemployed.Theyconsidertheequilibriumreaction

HOI+I−+H+↔I2+H2O
andverticaldiffusionofiodinespeciesthroughbrinechannelstotheseaicesurface,fromwhere
I2maybereleased.Theyconcludefromtheircalculationsthattheassumedemissionscanex-
plainobservationsofIOinAntarcticspring.Theproposedreleasemechanismis,however,still
controversiallydiscussed(Sander,2008;Carpenter,2008).
Inthefollowingstudy,theemissionratesoforganicprecursorsubstancesarekeptconstant
withineachnewmodelrun(similartothestudiesbyVogtetal.,1999)insteadoftheprecursor
concentrations,andtheresultingIOmixingratiosareinvestigated.Firstofall,theappliedmodel
isnowintroduced.

146

5.2DescriptionoftheCAABA/MECCAmodel

5.2DescriptionoftheCAABA/MECCAmodel

Theappliedatmosphericchemistrymodelhasanunderlyingmodularstructure-thecompletecode
isbuildupbyseparateparts,eachpartcoveringacertainaspect.

structureMESSyunderlyingTheThelinkbetweenthemodelparts,forexamplethetransferofparametersandresultsbetween
themodelpartsaswellastheorganisationoftheflowofvariablesisprovidedbytheMESSy
(ModularEarthSubmodelSystem)interface(Jöckeletal.,2005).MESSyisanewapproachtoan
EarthSystemModel(ESM).AnESMisconcernedwiththeentireEarthsystemandtheaimisto
combineallparts(suchastheatmosphere,biosphere,hydrosphereandothers)ofthesystemand
tocapturenotonlythesubsystemsthemselvesbutmostimportantlythefeedbackmechanismsand
interactionsbetweenthem.Earthsystemmodellingcanbeachievedbylinkingexistingdomain-
specificmodelstoeachother.Inthisdomainorientedapproach,singlemodelsforcertaindomains
suchasthewholeatmosphereortheentireoceansystemareusedandanadditionalprogram,a
so-calledcoupler,combinesalldomainsbyexchangingparametersandvaluesbetweenthem.
MESSyisratherprocessoriented.Inthisapproach,abasemodelisusedtocontrolthemodelruns
andcontainsforexamplethecentralmodeltimer.Allimportantprocessesaredescribedwithin
individualmodules(submodels)whicharedirectlylinkedtothebasemodelandarenotcombined
intodomainspecificmodelsfirst.Allsubmodelscanbeswitchonoroffseparatelyandacommon
interfaceisusedforthetransferofvariablesbetweenthebasemodelandthesubmodels.
TheMESSyinterfacestructurecanalsobeusedifnotacompleteESMshallberunbutacertain
subsystem.Inthepresentcasetheincludedsystemcomprisesmainlytheatmosphere,butalso
biogenicemissionsandsurfaceeffectsthroughdeposition.

dulemoMECCATheTheatmosphericchemistryisincorporatedintheMECCAmodule(ModuleEfficientlyCalculating
theChemistryoftheAtmosphere),andMECCAthenactsasasubmodelwithinMESSy.Thispart
containsallrelevantchemicalreactionsandtransformationsinthegasandaqueousphasesandthe
chemicalreactionkinetics(Sanderetal.,2005).Theinvolvedspeciesandtheirinitialvaluesare
defined.Forthesetupofthereactionschemes,theKineticPreProcessor(KPP)softwareisused,a
freesoftwarespecificallywrittenforthecomputationofchemicalreactionkinetics(Damianetal.,
2002;SanduandSander,2006).KPPgeneratesthemathematicalprogramcodeandperformsthe
integrationofthechemicalequations.
Thechemicalreactionsaremostlydescribedbyasetofstiffordinarydifferentialequations(ODEs).
Differentialequationsaredescribedasstiff,ifslowlyvaryingcomponentsneedtobeconsidered
alongsideotherswhicharequicklydamped,i.e.,incaseswhenanequilibriumisquicklyobtained.
Inthepresentcasequickandslowreactionsparticipateinthereactionkinetics.Forthenumerical
integrationofthedifferentialequations,severalintegratorsareavailablewithinKPP.Duetogood
stabilityproperties,allpossibleoptionsaresuitablefortheintegrationofstiffdifferentialequations
(SanduandSander,2006).Inthepresentstudy,therecommendedpositivedefiniteRosenbrock

147

5ModelingofatmosphericIOwiththeCAABA/MECCAcode

integratorisappliedasdefault.
Thedifferentialequationsdescribethetemporalevolutionoftheinvolvedchemicalspecies.Species
withvariableaswellasfixedconcentrationsareconsidered.TheJacobianmatrixJf(i,j)play
animportantroleintheintegrationofchemicalreactions.Itdescribeshowtheconcentrationof
thevariablespeciesevolvesintimedependingontheconcentrationsofallotherspecies(fixedand
variable).Thefunctionsfidescribethetimederivativesoftheconcentrationsdci/dtindependence
oftheothertracegases.Anexampleequationforaconcentrationcoftracegasa,whichdepends
on,e.g.,2othertracegaseshasthefollowingform:
fa=ddcta=−k1·ca·cb+k2·cc·cd,
wherethekjarethereactionratecoefficientsandca...cdaretheconcentrationsofinvolvedspecies.
Thefirsttermwithnegativesigndescribesalossreactionwhilethesecondsummandspecifiesa
productionprocess.Theequationforthetemporalevolutionofconcentrationscanhaveinrealitya
lotmoreterms,butthegeneralformcorrespondstotheaboveequation.Additionallossprocessesare
depositionandphotolysis.Thesedonotdependontheconcentrationsoftheothertracegases.The
Jacobianmatrixcontainsthepartialderivativesofthefunctionsfwithrespecttotheconcentrations
substances.allofJf(i,j)=∂fi
∂cjTheMECCAcodeincreasestheefficiencyoftheintegrationbynotusingthefullJacobianbut
savingonlythenon-zerovaluesandtheirrespectivematrixcoordinates,i.e.theJacobiansparsity
matrix.

Consideringtheimplementedchemistry,MECCAoffersthechoicebetweenmanydifferentscenarios.
Thetroposphericaswellasstratosphericchemistryarecoveredanddifferentchoicesfortheincluded
substancesarepossible.Dependingontheindividualrequirements,themodelusermayselectthe
appropriatescenario.Gasphaseaswellasaqueousphasespeciesandreactionscanbeincluded.
Ifnecessary,thechemicalmechanismcanalsobeextended,e.g.ifnewresultsfromlaboratory
studiesbecomeavailable.Thechemistryinvolvedisbasedonseveralpreviousstudies,specifically
thehalogenchemistryisimplementedfromSanderandCrutzen(1996)andvonGlasowetal.(2002),
withpartlyupdatedkineticsdata.Chemicalkineticsdataisavailable,e.g.,fromAtkinsonetal.
(2004),Atkinsonetal.(2006)andSanderetal.(2006b).Asummaryoftheimplementedchemistry
andkineticsdataisprovided(Sanderetal.,2005,supplementarymaterial1).

TheCAABAboxmodelusedasbasemodel
MECCAneedstobelinkedtoabasemodelprovidingtheatmosphericsituationforwhichcalcula-
tionsshallbeperformed.Thebasemodelcanbeofdifferentdegreesofcomplexity.Inthepresent
study,aboxmodelischosentoprovidethebasicsituation,butalsoacompleteGeneralCirculation
Model(GCM)canbeapplied.Forthecalculationoftheabundancesofiodinecompoundsunder
1http://www.atmos-chem-phys.net/5/445/2005/acp-5-445-2005-supplement.zip

148

5.2DescriptionoftheCAABA/MECCAmodel

definedconditions,azerodimensionalboxmodelissuitablehere.ThispartiscalledCAABA-
ChemistryasaBoxmodelApplication.Andtheboxmodelinconnectionwiththechemistrypro-
videdbyMECCAisreferredtoastheCAABA/MECCAcode(Sanderetal.,2009).Thescheme
inFig.5.1showsthegeneralstructureofthecombinedmodel.
ApartfromtheatmosphericchemistrywhichisprovidedbyMECCA,twoadditionalsubmodules
arerequired.TheSAPPHOmoduleprovidesthephotochemistry,i.e.photolysisrates.Emission
anddepositionreactionsandratesaregivenbytheSEMIDEP(SimpleEmissionandDeposition)
module.Withintheboxmodelstructureparametersforthegeolocationandtimearechosen,needed
forthecalculationofphotolysisrates.Meteorologicalsettingssuchasthetemperature,pressure
andhumidityaresetwhichinfluencethechemicalkinetics.Alsodifferentpropertiesoftheinvolved
aerosols(ortheoptionofexcludingaerosols)canbedefined.

elxmodA BoCAAB

Figure5.1:StructureoftheCAABA/MECCA
MESSy Interfacemodelingcode.Thelinkbetweentheindivid-
ualpurposesubmodelsisprovidedbytheMESSy
SAPPHOSEMIDEPMECCAinCAABAterface.servInesathesbasepresenmotdel.casetheboxmodel
photolysisemission, chemistry
tioniposde

Overviewoftheimplementediodinechemistry

ThecompletereactionmechanismincorporatedinMECCAcontainsabout200gasphasereactions
and60photolysisreactions,dependingonspecificchoices.The”baserun”inthisstudy,forexample,
has197gasphaseand64photolysisreactionswith112speciesinvolvedintotal.
TheiodinechemistrywhichisconsideredintheMECCAcodeisgraphicallydemonstratedin
Fig.5.2.Forthispurpose,allreactionsandconversionswhichinvolveiodinespecieshavebeen
selectedfromthecompletemechanismforanoverviewofthisspecificpartofthemodel.The
picturedoesnotgiveinformationabouttherelativeimportanceofthesinglereactions,butshows
whichspeciesareconnectedtoeachotherbychemicalconversions.Thedifferenttypesofboxesand
arrowschosenfortheindividualcasesareexplainedinthelegend.Theonlyspecieswhichisnot
achemicalmoleculeinthisfigureisIpart,whichsummarisesallparticulateiodinesubstances.The
differenttypesoffineparticlesthatcanbecreatedbyiodinespeciesarenotindividuallytreated
inthismodel.Currentunderstandingisthattheformationoffineparticlesfromiodineoxidesis
connectedtotheOIOmolecule,whichthereforeisthelinktoparticulateiodineintheapplied
de.coSomeexistingreactionsarenotyetincorporatedinthemodel.Aslaboratorystudiesprogress,the
chemicalmechanismcanbeextendedandcompleted.Sofar,nonewreactionshavebeenadded.

149

5ModelingofatmosphericIOwiththeCAABA/MECCAcode

partI

23NOO3HI3OHOINNOOOI2
OI+ HOHCHOBr2BrNOHOI2OIOHOOO2 HCHO+
222ClO33CH3HNOCl+ONOOBrI32OClO2NOOBrCH
2NONO2OOBr+OBrO3
332OOHCHIl7ICHr3IBCIOH

2OIN2NO2HOO2H2O3OIN3NO

2HOI

IOH

HIClI2CH

2I2CH

s)dacioxidesiodineanddineoIeniiodwithecies spnelogHapeciessodineiOrganic(e.g. speciesOther iodine

Photolysis reactionssctionlf-reaeSofReactionsspeciesiodinesubstancesotherofConversions

Figure5.2:ExcerptoftheMECCAchemicalmechanismshowingallreactionsandconversions
whichinvolveiodinespecies.Differentkindsofspeciesaremarkedbyindividualboxesanddifferent
classesofreactionsbyindividualarrowtypes.Ipartdenotescollectivelyallparticulateiodine
substances.Asprefactorsareomittedforclarity,theyhavetobeaddedwhereappropriatetomake
thestoichiometrycorrect.GenerallypresentspeciesasO2maybeomittedinsomecases.

150

5.2DescriptionoftheCAABA/MECCAmodel

+H

OHI3

IOH+HIO3
I2IO-IO3IBrI2H+
-YIBr-HOBrH2O2HOY
IClBr-ClOH-
IClHOClI-IO-
2+HHIY-H+O3H+
IY2-I2OH-HOYY-H-+
ClIOY-IYH+Y-Cl2+
HHOI+HHOIHO2HNOOIOHONO3INO3

onsHeterogeneous reactireactionsAqeous phase onsibrium reactiEquil

OIN2

IO22

Figure5.3:SchemeoftheaqueousphasereactionsrelatedtoiodineasconsideredbytheMECCA
model.Substancesoutsidetheschematicdroplet/liquidwatercontainingaerosolbelongtothe
gasphase(cp.Fig.5.2),allsubstancesinside(surroundedbycircles)aredissolvedintheliquid
phase.ThespeciesYdenoteseitherClorBr.ThemeaningofcoloursandboxesfromFig.5.2is
tained.main

151

5ModelingofatmosphericIOwiththeCAABA/MECCAcode

5.3Objectivesandmodelsettings
Thesatellitemeasurementspresentedinprevioussectionaddnewinformationtotheknowledge
aboutamountsanddistributionofIOwhichisimportantforthepictureofiodinechemistryand
itsrelevanceforatmosphericcomposition.RegionswithenhancedIOamountsaskforfurther
terpretation.inandexplanationInadditiontotheobservedabsoluteIOcolumnsanddistributionsaswellasupperlimits
forcertainlocations,itisofinteresttoestimatewhichsourcestrengthsofiodineprecursorsare
necessarytoproducetheobservedIOamounts.Thisneedstobecomputedtakingintoaccountup
todateatmosphericchemistryincludingiodinechemistryasfarasithasbeenassessedbylaboratory
efore.bstudiesIntheconductedmodelcalculations,firstofallrealisticscenariosoftheAntarctictroposphere
aresetup.TheaimisthentocomputeIOamountswhichareinaccordancewiththesatellite
observations.Asthequantitywhichiscomputedandusedbytheappliedmodelundergiven
conditionsisthevolumemixingratioofthetracegases,thecolumnamountsfromthesatellite
retrievalsneedtobetransformedbeforecomparisonispossible.AsdiscussedinSec.2.4,the
conversionfromtracegascolumnstoaconcentrationatacertainaltituderequirestheknowledgeof
thealtitudeprofile.AsthisisnotwellknowninthecaseofIO,theVMRinferredfromthesatellite
datashowsoneorderofmagnitudedifferencebetweena1kmboxprofileanda100mboxprofile,
forexample.Seasonalaveragesfortheslantcolumnslieintherangeof8×1012molec/cm2inthe
SouthPolarRegion.ForanAMFof4andthetwoboxprofilecasestheamountsoftheinferred
VMRliebetween0.8and8ppt(cp.Chapter3).Inmeasurementsfromtheground,severalpptof
IOhavebeenreportedforsomefavourableconditionsinAntarctica(Saiz-Lopezetal.,2007b),in
agreementwithaboveestimations.Theopenquestionatthispointis,whichconditionsareneeded
toproducetheseconcentrationsofIOunderAntarcticconditions.Thisshallbeinvestigatedinthe
followingbyusingappropriatesettingsandvaryingtheemissionsofprecursors.
ThelocationchosenforthemodelstudiesisHalleyStationclosetotheAntarcticcoast(75.5◦S,
26.5◦W).Forthislocation,thetypicalmeteorologicalconditions(temperature,pressure,humidity)
needtobefedintothemodelcode.Duetothevicinitytotheoceanfromthechosenlocation,the
chemistryofamarineboundarylayerisanappropriatechoice.Somebasicmodelsettingsthatwere
keptfixedforallmodelrunsarelistedinTable5.1.
Onecombinationofreasonableparametersandsettingsischosenasthe”baserun”towhichsub-
sequentmodelrunswillbecompared.Aselectionofimportantspecificsettingsissummarisedin
5.2.ab.TDifferentcombinationsofsettingshavebeentestedinordertogaininsightintowhichparameters
arethemostinfluentialfortheresultingconcentrationofIO.AllpropertieslistedinTab.5.2for
thebaserunhavebeenvaried.Onefocusisontheeffectofdifferentemissionrates.Thechosen
casesofemissionratesarerelatedtopublishedobservations,measurementsandothermodelstudies.
Usually,emissionratesaregiveneitherinunitsof”moleculespercm2persecond”orin”nmolper
m2perday”.Theconversionbetweenthetwooptionsis:

152

1·106molec/cm2/s=1.44nmol/m2/day.

SettingarameterPStationHalleycationLo◦S75.5Latitude◦W26.5LongitudeahP980Pressure80%yHumiditModelStartTime1stSeptember
ModelDuration15days
ModelTimeStep20minutes

resultsdelMo5.4

Table5.1:Settingsforthebasicconditionsvalidforallmodelruns.

SettingyertPropK253eratureempTAerosols2phases:sulphateaerosol,seasaltaerosol
EmissionsCH3Iat6·106molec/cm2/s(Vogtetal.,1999)
CH2I2at1.32·106molec/cm2/s(Carpenteretal.,2007)
5.3ab.Tcp.InitialamountsonlyCH3Iat2ppt
Table5.2:Modelparametersandsettingsasusedinthebaserun.

Whenevertheemissionrateisgiveninnmol/m2/dayintheliterature,thenumberisconverted
tomolec/cm2/sforconsistentuseanddirectcomparisonthroughoutthischapter.Table5.3lists
thedefinitionsoftheemissionsettingsappliedinsomeofthetestruns.Theemissionssetinthe
baserunaretakenfromtheonlyavailablemeasurementsoforganicprecursorfluxesconducted
intheAntarctic(Carpenteretal.,2007).Thetimeandexactlocationarenotcoincidingwith
highestsatelliteIOvalues,sothatmodeledIOisnotexpectedtobeatmaximumforthebaserun.
Otherobservationsandstudiesreporthigheremissionrates,buttheknowledgeofemissionrates
aresparse.ThebaserunresultsintoosmallIOamountstoexplainthesatelliteobservationsas
w.elobwnsho

resultsdelMo5.4

TheresultingIOamountsfromdifferentmodelrunsarenowinvestigated.Eachmodelrunproduces
timeseriesforeachincludedtracegasandaqueousphasespecies.Figure5.4displaysthetemporal
evolutionoftheIOconcentrationcalculatedbytheBaseRun.Forthefirstmodeldays,thetracegas
amountschangestronglyasthesituationismostlynotinequilibrium.Afterstabilityisachieved,
thedailymaximumIOamountrangesaround0.012ppt.Thisisfarbelowtheconvertedsatellite
results,about2ordersofmagnitudetoosmall.Oneinterestingfindingisthatthetimeofdaywith
maximumIOamountsapproximatelycoincideswiththeSciamachyoverpasstime,alsoforthe
followingmodelruns.Thisisaresultoftheongoingphotochemistrywhichinfirstplacereleases
iodinefromthephotolabileiodocarbons,butalsophotolyticallyreducesIOamounts.

153

5ModelingofatmosphericIOwiththeCAABA/MECCAcode

elLabRunBaseevemis2aemis2maxemis2MHogtemisV

MoleculeSettings

RunBaseemis2avetwoemittedcompounds
CH3I6·106molec/cm2/s(Vogtetal.,1999)
CH2I21.3·106molec/cm2/s(Carpenteretal.,2007)
emis2maxtwoemittedcompounds
CH3I6·106molec/cm2/s(Vogtetal.,1999)
CH2I24.2·106molec/cm2/s(Carpenteretal.,2007)
emis2MHtwoemittedcompoundsonly
accordingtomeasurementsatMaceHead(Carpenteretal.,2001)
CH3I6·106molec/cm2/s(Vogtetal.,1999)
CH2I21.4·109molec/cm2/s(Carpenteretal.,2001)
emisVogtallemissionsasin(Vogtetal.,1999)
CH3I6·106molec/cm2/s
CH3H7I1·107molec/cm2/s
CH2I23·107molec/cm2/s
CH2ClI2·107molec/cm2/s
emisCarpenterpersonalcommunicationofmodelauthorswithLucyCarpenter
CH3I6·106molec/cm2/s(Vogtetal.,1999)
CH3H7I1·107molec/cm2/s(Vogtetal.,1999)
CH2I23.8·108molec/cm2/s
CH2ClI1.3·109molec/cm2/s

Table5.3:Modelsettingsfordifferentprecursoremissionrates,accordingtodifferentobservations
publications.and

timeFigureperio5.4:dofMo15ddelledays.volumemixingratiosofIOforthebaserunshownforthecompletemodel

154

Resultsfromdifferentemissionscenarios

resultsdelMo5.4

Inanextstep,thebiogeniciodineprecursoremissionsasreportedintheliteraturearevariedas
giveninTab.5.3).TheresultingIOvolumemixingratioscomputedfromthesefivescenariosare
comparedinFig.5.5.Foreachcase,themaximumIOonthelastmodeldayisshown.Thebaserun
andtwootherscenarios(emis2maxandemisVogt)showvalueswellbelow0.8pptandconsequently
cannotexplaintheIOamountsretrievedfromsatelliteobservations.Theemissionsmeasuredby
Carpenteretal.(2001)inMaceHead(emis2MH)andthecombinationofemissionsreportedby
Vogtetal.(1999)andfrompersonalcommunicationofmodelauthorswithL.Carpenterleadto
IOVMRamountsbetween1and3ppt.ForbothcaseswithlargeIOamounts,theemissionof
polyhalomethanes(CH2I2andCH2ICl)arethecrucialsettings.Intheotherstudiesandscenarios,
thesewereeithernotpresentormuchlower.
Therefore,acomparisonofmodelledIOconcentrationindirectdependenceoftheemission
rateshasbeenperformed.Figure5.6containsinputfromseveralmodelcalculations-eachpoint
revealstheresultfromoneindividualmodelrun.ThemaximumIOVMRresultingforthelast
dayofthemodelruntime(15thSeptember)isshownversustheappointedemissionrateofmethyl
iodide(red)anddiiodomethane(blue).Thedatapointscorrespondtothemodeloutputvalues,
whilethesolidlinesare4thorderpolynomialfitsincludedinthegraphasaguidetotheeye.Very
clearly,theachievedIOamountsaremuchhigherforCH2I2emissionsthanforCH3I.Thisdifference
isnotonlycausedbythetwoiodineatomspermoleculeofCH2I2,butmainlybyitsmuchlarger
rate.photolysisTheabovefindingsshowthatthebiogenicemissionsobservedbyCarpenteretal.(2007)inAntarc-
ticaduringoneresearchcruisearenothighenoughtoaccountfortheIOconcentrationsobserved
inthepresentstudyatleastwiththecurrentchemicalscheme.Theconditionsduringtheresearch
cruiseprobablydidnotsupportthemaximumemissionspossible,timeswithhigherfluxesareex-
pected.AfewpointmeasurementswithinafewdaysinthemonthofDecemberweretakenandthis
cannotsufficetoexplainobservationsinallseasons.Atothertimesemissionsmaywellbemuch
higherandbiogenicpathwaysarestillconsideredrelevantforiodinereleaseintheAntarctic.In
thissense,themodelresultsdonotcontradicttheassumptionofbiologicalrelease.

delledMo5.5:Figurevolumemixingratiosof
IO(themaximumofthe
lastmodelday)from
scenariostdifferenefivemissionindividualwithinstatedassettingsTab.5.3.Modella-
bels:1)baserun,2)
4)emis2max,emisVogt,3)5)emis2MH,emis-
.terenCarp

155

5ModelingofatmosphericIOwiththeCAABA/MECCAcode

thcodeFigureindep5.6:endenceMaximofumthedailyappliedIOvaluesemissionontherates15ofmoCH3delIday(red)andcalculatedCH2bI2ythe(blue).CAABA/MECCA

Emissionsreportedatothersitesandfromotherstudies,leadtomodelledIOamountsinthe
rangecalculatedfromthesatelliteresultsinagreementwithground-basedmeasurements.With
theseemissionsettingschosen,IOmixingratiosontheorderofafewpptareachieved.
InordertoreachtheIOamountsseenfromsatellite,emissionsofCH2I2aremostimportant.
ObservedamountsandemissionratesofmethyliodideonlyleadtoanIOmixingratioonthe
orderof0.01pptandcannotexplaincurrentsatelliteobservationsifIOisconfinedtothelower
atmosphericlayersaswasreportedbySaiz-Lopezetal.(2007c).
Evenifhighbiogenicemissionsareassumed,theresultingIOamountsareratheronthelow
side.Owingtothelownumberofavailablefluxmeasurements,itispossiblethathigheremission
ratesoccurwithouthavingbeenobservedyet.Especially,nomeasurementswereperformedwithin
theAntarcticseaiceduringtheSpringtimeperiod.Inadditionitseemstobelikely,thatsome
potentiallyimportantprocessesarenotyetconsideredinthemodel.

5.5Proposedmodelextensions

Asmentionedabove,appliedatmosphericmodelsmaybeincomplete,possiblyinthepresentcase,
somereactionpathwaysareneglectedwhichhaveasignificantinfluenceontheresultsornewkinetic
studiesmayrevisetheknowledgeoncertainreactionrates.Thepresentedcalculationsresultin
IOamountswhichareclearlyonthelowsideofobservations.Inregardofpotentiallymissing
chemistryorinappropriatekineticsdata,candidatesforfuturemodelchangesorextensionsneedto
tified.ideneb

156

5.5Proposedmodelextensions

Onepossiblyimportantreactionhasbeenidentifiedtobeneglectedinthemodel,whichisthe
photolysisoftheOIOmolecule.Thisreactionhasbeencontroversiallydiscussedintheliterature.
Thequantumyieldoftherespectiveproductsisnotagreedon.Recentnewlaboratorywork
(GómezMartínetal.,2009)givesnewevidencethatOIOphotolysiswithaquantumefficiency
ofunity,whileformerlaboratorystudieshadreportedaquantumyield0.05(Tuccerietal.,2006).
GómezMartínetal.(2009)estimatethatiodineemissionsneedtobefivetimeslargerincasethe
quantumyieldtakesonthesmallervalueof0.05insteadof1.Thefactthatonemissingreaction
whichispotentiallyimportantmaychangethemodelledIOamounteffectivelyfivefoldshowshow
importantthecompletenessofthechemicalmechanismsistocomputevaluesinagreementwith
realworldobservations.TheeffectofincludingtheOIOphotolyseswithaquantumyieldof1for
iodineatomswillbeinvestigatedinfuturestudies.
Inaddition,severalheterogeneousphasereactions,recyclingonsnow/aerosolsurfacesneedto
beincluded.Currently,particulateiodineactsasasinkinthemodel,whileseasaltaerosolsare
modelledtoexchangesomecompoundsbetweenaqueousandgasphase.However,thereleaseof
gaseouscompoundsfromiodineparticlesmaywellbepossible.ConsideringthemoleculeI2O5,
whichisacompoundofparticulateiodineandofhygroscopicnature,thefollowingreactionsmight
proceed,wheregasphasespeciesreactwithaqueousphasesubstancesontheaerosolsurface:

I2O5+H2O(aerosol)→2HIO3(aerosol)
OH+HIO3(aerosol)→H2O(aerosol)+IO3(g)
IO3+hν→OIO+O
O+IO2→Inthiscase,gasphasereactiveiodinewouldberecovered.Presently,itisnotknown,ifthesespecific
reactionsoccur.Reactioncyclesofthistype,involvinggasphaseandaqueousphasespeciesatthe
sametime,deservefurtherattentionandmightimprovetheunderstandingofiodinerecyclingfrom
phase.particulatethe

157

6Analysingshipbornedataforthe
improvementofDOASretrievals

InApril-May2008,aBremenMAX-DOASinstrument(Sec.1.9.2)wastakenontoacruisethrough
theAtlanticoceanontheresearchvesselPolarstern.Theinstrumentisdesignedforobservationsin
theUVandthevisiblewavelengthregions.Inordertoinvestigatethewaterleavingradiance,data
fromthiscruiseisanalysedanddiscussedhere.Theaimforthisstudyistoextractacorrection
spectrumfortheimprovementofsatelliteretrievalsoverwaterbodies.

6.1Motivationforthefollowinganalysis

AsdiscussedinSection3.1,thereareregularartefactsinthesatelliteretrievalswhicharecommonto
severalofthetracegasproductsdeducedfrom,e.g.,Sciamachyobservations.Oneveryprominent
artefactthathasalreadybeenexaminedfromdifferentsidesarethenegativetracegasamounts
abovetheclearoceanregions.Itisnotclearuptonow,whyinthesewidespreadregionssome
tracegasproductsshowsomenegativevalues,e.g.IO(Schönhardtetal.,2008)orCHOCHO
(Wittrock,2006).Inotherproducts,nonegativevaluesappear,butthepatternofthedifferent
oceanregionsneverthelessshowsupintheglobalmaps.Theclearoceanregionscontainlow
chlorophyllconcentrationandlowbiologicalactivity.
Theproblematiceffectonthespectroscopicmeasurementsmighthavedifferentsources.The
sunlighttravelsalargerpathlengththroughclearwater,wherethelightpathisnotobstructedby,
e.g.,largesuspendedparticles.Thewatermoleculesthemselvesandnarrowbandspectralfeatures
mayaffectthespectrumofthetransmittedlightstronger.Theseinfluenceshavetheirorigininab-
sorption,butalsoinscatteringprocesses.Theabsorptionbyliquidwaterisbroad-bandandrather
weakatvisiblewavelengths,butmayaffectatmosphericmeasurementsnevertheless.Itsabsorption
coefficienthasbeenmeasuredinthelaboratory,e.g.byPopeandFry(1997).Otherfeaturesare
lesswellknownandmighthaveaconsiderableimpact,likethespectrallymorestructuredabsorp-
tioncoefficientsbyChlorophyll,certainPhytoplanktontypesorCDOM(ChromophoricDissolved
OrganicMatter).However,theclearwaterregionscontainjustlittleofthesesubstances,whichare
moreprominentinthebiologicallyactiveareas.
Clearoceanregionsoftenarethecalmerregions,leadingtoatendencytowardsaflattersurface.
Therefore,polarisationeffectswillplayalargerrolehere.Anothereffectistheinelasticscattering
ofphotonsatwatermolecules(andothermoleculeswithintheoceanwater),i.e.thevibrational
Ramanscattering.Themoleculesareexcited,e.g.,fromthegroundleveltoanelectronicallyhigher
levelandrelaxbacktoavibrationallyexcitedstateoftheelectronicgroundlevel.Thisresultsina
wavelengthshiftoftheemittedphotonascomparedtotheincidentphotonandbyexactlythesame

159

6AnalysingshipbornedatafortheimprovementofDOASretrievals

principleasdiscussedfortheRingeffect(cp.Sec.1.6.3)causesanin-fillingofstrongabsorption
structures,especiallyoftheFraunhoferlines.CalculatingthespectraleffectofVRShasalready
broughtsomeinsightintothisprocessandtheinfluenceontracegasretrievals(Vasilkovetal.,
2002;Vountasetal.,2003).Nevertheless,someinsufficientcorrectionorsomeadditionalfeatures
intheseclearoceanregionsarestillapparent.
Insteadof(oradditionallyto)calculatingtheeffectsthatwaterhasonthelightspectrum,it
shouldbeusefultomeasuretheseeffectsdirectly.Therefore,theideadevelopedtouseaDOAS
systemthatwouldspecificallyobservethewaterleavingradianceonthePolarsterncruise.Forthis
purpose,someadditionalviewingdirectionswereaddedtotheusualcycleofmeasurementscans
throughtheatmosphere.Additionalanglesataline-of-sight(LOS)below0◦(belowthehorizon)
werechosenwheretheinstrumentislookingdownwards-towardstheoceanwater.

6.2detailsmeasurementandInstruments

ThetrackoftheshipcruiseduringwhichtheDOASmeasurementswererecordedisshowninFig.6.1
andtookplacefromApril18th-May20th2008.Theusualspeedoftheresearchshipwasclose
to11kn(i.e.20km/h)withdailystationperiodswithoutshipmotionatnoonorintheafternoon
foraround1to2hours.Thesecalmstationperiodswerequiteimportantformeasurementsinlow
viewingangles,asthebowwaveortheassociatedseasprayduringusualmotionpartlyobstructed
thesightintotheoceanwateratanglesbelowthehorizon.

fromFigurePun6.1:taTheArenastrack(Chile)ofthetoBremerhaANT-XXIV/4venP(Germanolarsterny).cruise

TheDOASinstrumentonboardcomprisestwospectrometerunits,onemeasuringintheUVand
oneinthevisiblewavelengthregion,whichwerebothservedwithlightbyonemutualtelescope.The
viewingangleisdeterminedbythepositionofamirrormountedwithinthetelescopebox.Further
basicinformationontheBremenMAX-DOASsystemsisgiveninSec.1.9.2.Table6.1contains
somedetailsontheinstrumentparameters,whileFig.6.2showsaschematicoftheinstrument’s
ship.theongeometryviewing

160

6.2Instrumentsandmeasurementdetails

InstrumentDataUVinstrumentVISinstrument
SpectrometerActon500Acton275
lines/mm300lines/mm600GratingWavelengthRange327-410nm400-710nm
MeasurementInformationUsualValue
typicalexposuretime0.5s
integrationtime(zenith)120s
integrationtime(otherdirections)60s

Table6.1:Overviewoftheinstruments(top)appliedonthePolarsternshipcruise,aswellassome
relevantmeasurementsettings(bottom).

DirectionAZenith
escopetelanSc=0

BectionDir

shipFigure6.2:Sketchoftheviewinggeometry
oftheDOAStelescopeonboardtheresearch
vessel.Thesequenceoftheviewingangleβ,
alsoreferredtoaslineofsight(LOS),wascho-
sensuchthat◦viewinginto◦theoceanwaterwas
oceanusualfacilitatedscans(−ab90ove<βthe<0horizon)in(β>addition0◦).tothe

anglesviewingofSequenceTheusualmeasurementsequencecomprisesviewingtowardsthezenith,viewingatLOS=−60◦(into
thewater),atLOS=30◦(intotheatmosphere)andviewingatthescananglesbetweenLOS=−2◦
andLOS=14◦in2◦steps.Someadditionallowanglesbetween−60◦and−48◦throughalower
windowinthetelescopeboxandbetween−22◦and−2◦throughtheupperwindowareincluded
duringtheoccasionalstationperiodsoftheship(usuallyonceadayforabout1to2hours).
Theintegrationtimeisusually60sforallslantdirections,and120sforthezenithobservation.A
completesequencethereforetakesabout20minutes.Whenthesunistooclosetotheobservation
direction,theslantviewingdirectionsareinterruptedandonlyzenithobservationsareperformedto
avoidtoointenseilluminationandsaturationduetodirectsun.Atypicalmeasurementsequenceis
depictedinFig.6.3,wheretheLOSisplottedversusuniversaltimehereforthe10thofMay2008.

rrectioncoDataBeforethePolarsternDOASmeasurementscouldbeproperlyanalysedandinterpreted,somecor-
rectionshadtobeperformed.Owingtotheshipmotion(especiallythepitchandroleangles),the
viewingdirectiongivenbythetelescope’smirrorangleanddeterminedbythecomputercontrol
isnottherealviewingdirectionfromwherephotonsarecollected.Theusualintegrationtimeis

161

6AnalysingshipbornedatafortheimprovementofDOASretrievals

Figure6.3:Lineofsightversustime[UT]forthe10thofMay,atypicaldayoftheresearch
campaign.

60sor120s,whilethefrequencyoftheshipmotionliesontheorderof0.1Hz,dependingon
conditions.Inanycase,theintegratedobservationsthencontainmeasurementsindifferentviewing
directions,averagedoverawiderrangeofangles.Inordertoreceivemeasurementswithinasmaller
intervalofanglesandtobesureabouttherealviewingangle,theoriginalspectra(recordedat2Hz)
weresorted.Thesortingroutineneedstheinformationontheshipmotionasinputapartfromthe
originalspectra.Informationonthepitchandrollanglesisavailablefromtheship’snavigation
platform.Thesortingandaveragingroutineperformsthefollowingsteps:

•Reading-inofthepositioninformationfromtheshipinhightemporalresolution(10Hz).
•Reading-inoftheoriginalspectra(beforeintegrationover60sor120srespectively)withthe
givenmeasurementtimeandLOSinformationintheheader.

•TheLOSgivenineachspectrumiscorrectedbytheadditionalanglefromtheshipandis
assignedtoanewclassofmeasurements(in2◦steps).

6.3Retrievaloftheliquidwaterabsorption

Firstofalloneneedstoknow,iftheradiationthatwasrecordedinthewaterviewingdirection
hasreallypassedthroughtheoceanwater.Ifthisisthecase,thespectrashouldhavepickedup
someliquidwaterabsorption.Theabsoluteabsorptioncoefficientinthevisiblewavelengthregion
andthedifferentialabsorptionstructureinasmallerspectralintervalareshowninFig.6.4from
laboratorymeasurementsbyPopeandFry(1997).
Fortheretrievaloftheliquidwaterabsorption,awavelengthregionwaschosenthatstilloverlaps
withtheregionsoffurtherinterest(seebelow)andwherethewaterabsorptionhassufficientspectral
structures.Thefollowingretrievalswereperformedforthewavelengthregionfrom420to540nm,

162

(a)

(b)

6.3Retrievaloftheliquidwaterabsorption

Figure6.4:(a)AbsoluteabsorptioncoefficientofpureliquidwaterasmeasuredbyPopeandFry
(1997).(b)Excerptofthisspectrumindifferentialform,i.e.apolynomial(hereofdegree5)has
beensubtractedtoenhancethehigherfrequencystructures.

i.e.theregioncoveredinFig.6.4(b).TheDOASfitsettingsthatwereappliedfortheretrievalof
theliquidwaterabsorptionarelistedinTab.6.2.
retrievaterwLiquidalWavelengthregion420-540nm
gasesraceTHH22OO(g)(liq),296(PKopeand(HITRANFry,data1997)base)
O,241K(Burrowsetal.,1999a)
O34,296K(Greenblattetal.,1990)
Consideredeffects5thorderpolynomial
straRingyligheffectt(Scia(constanttran)offset)
Table6.2:DOASfitsettingsfortheretrievalofliquidwaterabsorptioninthevisiblewavelength
region.

Thefitfactorsfortheliquidwaterabsorptiongivetheaverageeffectivelightpaththroughthe
water,asinthiscase,nottheabsorptioncrosssectionbuttheabsorptioncoefficientisapplied.
AsamplefitresultforthefittingofthewaterabsorptionstructurescanbeinspectedinFig.6.5.
Theexpectationnowis,thatforwaterviewinganglesapositivefitfactorshouldresultfromthe
fitandasmallornegativeoneforhigherviewingangles.Thebackgroundspectrumistakenfrom
the30◦-directionabovethewatersurface.(Thiswaspreferredincomparisonthethezenith-sky
spectrumtoavoidpotentialadditionalstructuresrelatedtothemirror.)Itispossiblethatthe
30◦-directionalsocontainsasmallsignatureofthewaterabsorptionfromwaterleavingradiation
scatteredintothefieldofviewoftheinstrument.Incaseoflesswatersignatureinthecompared
spectrum,theresultingfitfactorwouldbenegative.Thisisobservedinsomeoccasions,especially
inthefitofthezenithspectraagainstthe30◦-direction.InFig.6.6thewaterabsorptionfitfactor

163

6AnalysingshipbornedatafortheimprovementofDOASretrievals

(liquidwaterpath)isplottedversustheLOSfortwosampledays(May2ndand12th,2008).

Figure6.5:Samplefitresultforthewaterabsorptionstructuresshowingthedifferentialoptical
(cp.depthFig.vs.w6.4)avandelength.themTheeasuremenscaledtreference(red)waspstakectrumenat(blacak)LOSoforiginates-2◦andfromshoPwsopehereandtheFryfit(1997)result
ofthewaterabsorptionincludingtheresidual.

6.4Themixedwatereffect

Inthefollowing,remainingspectralstructurespossiblyhintingatunconsideredwatereffectsshall
beextractedfromthemeasurements.Forthispurpose,aDOASfithasbeenperformed.The
objectiveistoremoveallknownabsorptionlinesandspectralfeaturesfromthemeasuredspectra
andtoreceivearesidualthatcontainsinformationaboutallprocessesthatarestillunaccounted
for.Itisimportanttoexcludeasmanyotherfeaturesfromthespectraaspossibletoreducethe
probabilityofartefactsandremnantsofothertracespeciesintheresidual.Theremaindershould
berelatedtotheprocessesaffectingthelightintheoceanwater,whichofcourseneedstobetested.
Fortheextractionofthewatereffect,suitableDOASfitsettingsneedtobechosen.

Extractionofthewatereffect
Thewavelengthregionwhichisofspecialinteresthere,liesaroundthepossiblefittingwindowsof
IO,CHOCHOandNO2.Inthespectralintervalwherethesetracespeciesaretypicallyanalysed
thementionedproblems(cp.Sec.6.1)occur.Thefinalgoalistoimprovethesatelliteretrievals
fortheseatmosphericgases.Asitturnsoutattheendofthefollowingprocedures,thediscovered
spectrumforthemixedwatereffectdoesindeedpickupthesignaloverclearwaterregions.But
unfortunately,theobservedeffectisnotabletoeliminatethediscussedproblemsoverclearwater
regionsforthesetracegaseswhenincludedinthesatelliteretrievals.
Consideringthespectralregionofinterest,thepresentedanalysiswillconcentrateontheob-
servationsinthevisiblewavelengthregion.Thecentralwindowthatwaschosenfortheanalyses

164

(a)

(b)

6.4

The

aterwmixed

effect

Figure6.6:FitfactorsoftheliquidwaterabsorptionretrievedfromthePolarsternDOASmea-
surementsfortwodifferentdays,the12th(a)andthe2nd(b)ofMay,respectively.For(b)theLOS
wasrestrictedtothewaterviewingdirections.

165

6AnalysingshipbornedatafortheimprovementofDOASretrievals

liesbetween411and455nm,containingsomeprominentfeaturesofIO,CHOCHOandNO2.Itis
necessarythatthechosenwavelengthwindowextendsafewnmtobothwavelengthsidesbeyond
thelimitsoftheusualfittingwindowsofthesetracegasesbecausethefinalproductofthepresent
analysis-theresidualstructure-shallbeusedinsubsequentDOASretrievalsasreferencespec-
trum.Referencespectraalwaysneedtobeavailableinaspectrallywiderregionthanthefitting
window,astheyneedtobeconvolvedwiththeinstrument’sslitfunction,alsoatthebordersofthe
window.Additionally,someshiftandsqueezeisallowedforthereferencespectrum.Soaslightly
widerwindowthanseeminglynecessaryisselected.
TheappliedDOASretrievalsettingsfortheextractionofthemixedwatereffectfromthe
remainingresidualsaregiveninTab.6.3.Themeasurementsofinterestarethespectrarecordedat
lowelevationangles(negativeLOSθ)wherethelighthaspassedthroughtheoceanwaterbefore
ectrometer.sptheteringensettingsalretrievASDOwavelengthregion411-455nm
backgroundspectrumdirectionA(LOS=30◦)
polynomial5thorder
gasestraceincludedONO,2,241221KK(Burro(Burrowswsetetal.,al.,1999a)1998)
3O4,(Greenblattetal.,1990)
H2O(g),296K(HITRANdatabase)
effectsconsideredRingeffect(calculatedbySciatran)
straylight(constant)

Table6.3:RetrievalsettingsfortheDOASfitfromwhichtheresidualsareusedtoextractthe
effect.aterwmixed

Tomakesure,thatthefitisreasonablystable,retrievalsinshiftedorextendedsurrounding
wavelengthregionswereperformed.Onlyminorchangesintheresidualwithinthecentralrangeof
thewavelengthwindowwereobtained,sothattheDOASretrievalcanbeconsideredstableforthe
settings.hosencForthechoiceofthepolynomialdegree,thestabilityoftheresidualwastestedforanincreasing
polynomialdegree.Finally,thesmallestdegreewaschosen,fromwhichonnolargeeffectswere
encounteredwhenthedegreewasincreased.Forthegivenmeasurementsandretrievalsettings,this
wasapolynomialof5thorder(sixdegreesoffreedom).
Fortheextractionofthemixedwatereffect,onlythoseresidualsfrommeasurementsduring
shipstationwereselected.Asdiscussedandshownintheprevioussection,fortheseobservationsthe
recordedlightreallypassedthroughtheoceanwaterbeforeenteringthespectrometerandshould
thereforecontaintheunaccountedwatereffects.AftertheDOASretrievalwithabovesettings,
thefitresidualsoftheobservationsduringstationperiodwereextractedandaveraged.Forthis
purpose,onlymeasurementsbetween-60◦and-50◦wereselected.FortheindividualLOS,different
numbersofspectraareavailable,becauseobservationswerenotperformedforallviewingangleson

166

6.4Themixedwatereffect

everydayandnotalwaysduringstationperiod.OnlyinB-direction(i.e.at-60◦LOS,cp.Sec.6.2)
regularmeasurementswererecordednearlyeverydayalsoduringthestationperiods.Theanalysis
isrestrictedtomeasurementsatstationtimes,becausetheinfluencefromspraywaterduringship
motionwasfoundtoprohibittheclearsightintotheoceanwater.Intotal,around100spectra
couldbeusedforaveraging:
•FromthelowLOSscans(between-58◦and-50◦):
3daysfor32◦,13daysfor-56◦,3daysfor-54◦,13daysfor-52◦,1dayfor-50◦withoneor
twomeasurementsaday.
•FromthelowestLOS,theregularB-direction(at-60◦):
19daysnearlythroughoutthecampaignwithatotalof56singlemeasurements.
InFig.6.7,theresidualspectraforthedifferentviewinganglesareshown.Alloftheresidu-
alsoriginatingfromtheabovementionedmeasurementswereaveragedtoyieldaweightedmean
residual.ThisisalsoincludedinFig.6.7ingrey.Theplotshows,thattheresidualsfromthe
differentlowviewingangleshaveseveralstructuresincommon.Somenoiseonthespectraand
individualfeaturesinsinglemeasurementscertainlyleadtosomedifferencesbetweentheshown
residuals.Butnevertheless,atseveralspectralpositions,thereareclearsystematicfeatures.Here,
attentionshouldbeturnedtothestructuresaround424-428nm,434-438nm,and443-446nm.
Additionallytothegeneralsimilaritybetweentheresiduals,inthesepositionspersistentstructures
t.apparenmostare

Figuremeasuremen6.7:tsAtakveneragedduringresidualshipspstation.ectraforThewdiffereneighttedaLOSverage(shoownverinallLOSdifferenistshowncolours)ingreyfrom.the

Theextractedresidualspectrumexhibitsfairlyhighfrequentstructures.Presently,itisnot
possibletoclearlylinkthesedetectedsystematicstructureswithacertainphysicalorotherprocess.
Consideringtheinputdata(Fig.6.7),itbecomesclearthatapartfromthesystematicfeatures,

167

6AnalysingshipbornedatafortheimprovementofDOASretrievals

whicharecommontoallresidualspectra,alsoother,lessspecificpeaksandstructuresexist.In
theseotherregions,thedifferencesbetweentheindividualmeasurementsaremoreapparentand
certainly,thereissomeremainingnoiseintheresults.Soalsotheaveragedresidualspectrum,
assignedtothemixedwatereffecthere,containssomeyetundefinedamountofnoise,instrumental
effectsandretrievalrelatedfeatures,whichcannotberemovedfromafinitesetofmeasurements
recordedundervaryingatmosphericconditions.
Itwouldbeinterestingandinsightfultodeterminetheoriginofeachofthedetectedpersistent
features,butthesimilaritytoknownspectraleffectsliketheVRSortheabsorptionbychlorophyll
isnotevident.ThecorrelationcoefficientinthiswavelengthregionbetweentheVRSspectrum
andtheretrievedwatereffectyieldsavalueofc=0.05,i.e.thetwospectraarenearlylinearly
independent.Aclearattributionisnotpossibleonthebasisofthecurrentknowledge.
Certainly,itisnecessarytofindout,ifamainportionoftheseextractedstructuresreallyis
connectedtoprocesseshappeningintheoceanwaterandarenotof,e.g.,instrumentalorigin.This
meansespecially,thatthestructuresshouldappearonlyinthewaterviewingLOSandnotinthe
zenithdirectionforexample.Thisisinvestigatedinthenextsection.

Fitfactorsofthemixedwatereffect
Aconfirmationisneededthatasubstantialpartoftheextractedaverageresidualisreallycaused
bymixedeffectsintheoceanwater.TheextractedwatereffectcanbeincludedinafurtherDOAS
fittofindouthowmuchofthesewatereffectfeaturesisfoundinwhichoftheavailableviewing
angles.Theabsolutevalueofthefitfactorisexpectedtobelargeforanglesthatviewintothewater
andclosetozeroforanglesviewingintotheatmosphere.Theextractedresidualisnowincluded
inaDOASfitinexactlythatspectralformasshowninFig.6.7.Thatmeans,thatabsorption
linesappearasminimainthisreferencespectrum.Thisisexactlycontrarytothealgebraicsignof
absorptioncrosssections,whereabsorptionlinesappearaspositivepeaks.Therefore,thefitfactor
willbenegativehereincasethestructuresofthedeclaredwatereffectaredetected.Consistently,
thesameDOASfitsettingsarechosenasbeforeinthewavelengthregionfrom411to455nm.
TheDOASretrievalofthemixedwatereffectisnowperformedforthemeasurementsofthe
entirecampaign.InFig.6.8,anexamplefitresultisshown,withthescaledwatereffectspectrum
(black)andthefitfromtheactualmeasurementcontainingtheremainingresidual(inred).The
(dimensionless)fitfactorforthisexamplefitis-1.01±0.10,implyingadetectionofthewatereffect,
i.e.withcorrectalgebraicsign.Thestrongerfeaturesfromthewatereffectspectrumarewellvisible
inthemeasurement(Fig.6.8).Additionalnoiseandunassignedstructuresareapparentalso,asthe
curveshowninredisanindividualmeasurement.
Nowthefitresultsshallbeanalysedfurtherbyexamininghowthewatereffectfitfactordepends
ontheviewingangle.TheresultsofseveraldaysofmeasurementsarecombinedinFig.6.9.The
respectivefitfactorsforthemixedwatereffectareplottedversusthelineofsight,firstofallonly
foranglestowardsthewater.Inthe-2◦-direction,alargevariabilitycanbeseen.Inthisdirection,
measurementswererecordedthroughouttheday,sothatseveraldifferentconditionsarecombined
here.Theaveragefitfactorresultingat-2◦-directionis0.17,butthescatteramountsto±0.43,
whichissubstantiallylarger.

168

6.4Themixedwatereffect

Figure6.8:FitresultshowinganexampleretrievalofthewatereffectfromonePolarsternmea-
surementonMay10th.Theblackcurveisthescaledwatereffectasitwasextractedfromthe
residuals.Themeasurementfitresultcontainingtheactualresidualisplottedinred.

TheexpectedvaluesforpositiveLOS(abovethewater)areclosetozero,whilesmalldeviations
fromzerocanbeunderstoodbecausealsoatviewingdirectionsabovewater,partsofthewater
leavingradiancemaybescatteredtowardstheinstrument.ForMay1st2008,theretrievalsin
zenith-skydirection,e.g.,yieldanaveragevalueofthewatereffectfitfactorof-0.30±0.38.This
valueisstillcompatiblewithanaverageofzero,butalsoexhibitsalargevariability.Thislarge
variabilityismostlyduetoasubstantialnoiseportionstillpresentintheextractedwatereffect
spectrum.Additionally,thewatereffectwasextractedfromspectrarecordedatquitedifferent
meteorologicalconditions.Itisnotpossibletofurtherhomogenisetheevaluationasthedataset
isalreadyratherrestricted.Nevertheless,theretrievalsshowthecorrecttendencyandbehaviour
withvaluesaroundzeroatviewingdirectionsabovethewaterandnegativefitfactorsforthelow
angles.Thefitofalinearfunctiontothewatereffectfitfactor(WEFF)withrespecttoLOS,based
onthedatainFig.6.9,resultsinthefollowingdependency:

WEFF=(0.023±0.005)/◦·LOS+(0.138±0.080)
Althoughthedatabasisisrathersmall,theexpectedbehaviourisrevealed.Astronglynegative
WEFFwouldberetrievedfordirectdownwardviewing(i.e.thecalculatedWEFFatLOS=-90◦)
asthewaterpathwouldbelong.Consideringthatthewatereffectspectrumwasretrievedat
viewinganglesbetween-60◦and-48◦,forwhichthefitfactorshouldbearound-1,afitfactorof
WEFF(-90◦)=−1.93±0.54seemsreasonablefordownwardviewingintothewater.Calculating
fromthistheinterceptwiththex-axisLOS0,givingtheLOSwherethewatereffectfitfactoris
zero,avalueofLOS0=-6.0◦−+36..19◦◦isobtained.Thisintervaldoesnotcontainthehorizon,butis

169

6AnalysingshipbornedatafortheimprovementofDOASretrievals

Figure6.9:Fitfactorsoftheextractedmixedwatereffectfordifferentlinesofsight.Forthelow
beanglesendetectedviewinginintothesewatervcases.aluesVaroundariabilities-1areoftheretrievfitedfactor,indicatingespeciallythatfortheamixedLOSofwater-2◦,effectarevehasry
large.

veryclosetothe0◦direction,andratherlargeuncertaintyrangesareobtainedduetothediscussed
reasons.

6.5Retrievalofthewatereffectinsatellitedata

Themainobjectiveoftheanalysisofthewatereffectistheimprovementofcertainsatelliteretrievals
includingtheIOretrieval.Forthispurpose,itneedstobepossibletoretrievefitfactorsforthe
extractedmixedwatereffectfromsatellitedata.Ifthedeclaredwatereffectactuallyoriginatesfrom
lightpassingthroughoceanwaterandiftheeffectisstrongenoughtobeperceivableinthesatellite
measurementsalso,itshouldbepossibletodistinguishclearoceanregionsfromturbidareasand
landinthesatelliteobservations.
Firstofall,thewatereffectisincludedinthestandardIOretrievalforSciamachyobservations.
Allotherparametersarekeptunaltered,onlythenewspectrumisconsideredadditionally.The
resultingfitfactorcanthenbeplottedonaglobalmap.Asanexample,theretrievalresultfora
threemonthperiodisshowninFig.6.11(a).
Thesameprinciplecanbeappliedforthedetectionofthewatereffectinothersatelliteobservations.
Inanothertest,thedeclaredwatereffectspectrumisincludedinafitoptimisedfortheretrieval
ofNO2fromthesatellitesensorGome-2(cp.Sec.1.9.3).Fig.6.11(b)containstheresultingfit
factorfromthisanalysis,averagedforonemonth(September2008).
Inbothretrievals,thepatternofdeepandclearwaterareasisproperlyidentified.Especiallythe

170

6.5Retrievalofthewatereffectinsatellitedata

Figure6.10:Mapshowingthecolourcodedfitfactorofliquidwaterabsorptionasretrievedfrom
GOME-2observations.Thefitfactordenotestheeffectiveliquidwaterpath(LH2Oliq)thephotons
takeonaverage.Theretrievedvaluesareaveragedforonemonth(September2008).Deepandclear
oceanregions(green-red)canbedistinguishedfromturbidwaters(e.g.duetobiologicalactivity)
andlandmasses(blueandpurple).TheGOME-2dataforthismaphavebeenkindlyprovidedby
AndreasRichter,IUPBremen.

algebraicsignisalsocorrect,asthenegativesign(bluecolour)impliesthedetectionofthewater
effect.Fitfactorswithnegativesignarefoundoverthedeepandclearoceanregions.When
comparingtheresultsofthewatereffectandtheliquidwaterpathfromsatellite(Fig.6.10)the
agreementofthepatterniswellrecognizable.
Sofar,theanalysishasbeensuccessful.Unfortunately,includingthewatereffectinthesatellite
retrievaldoesnotimprovethetracegasproductsunderinvestigation.TheIOretrievalwhichsuffers
fromnegativetracegasamountsretrievedovertheclearoceanregions,stillexhibitsthesameresults
evenafterincludingthecorrectionspectruminthesatelliteretrieval.
Astheaboveanalysisyieldspromisingresultsincapturingthewatereffectinthewaterviewing
directionsfromtheshipmeasurementsaswellasthedeepoceanpatternfromthesatellitemeasure-
ments,theoverallconceptseemstobesuitableforgaininginformationabouttheinfluenceofocean
waterontheobservedlightspectra.Nevertheless,thesatelliteretrievalscouldnotbeimprovedup
tonow.Alargerdatasetfromtheshipbornemeasurementswouldmaybeimprovethesituation
asthequalityofthecorrectionspectrumisnotveryhigh.Possibly,theremainingscatteringand
someresidualfeaturesnotrelatedtowaterinfluencesinhibitaperceptibleimprovementofsatellite
als.retriev

171

6AnalysingshipbornedatafortheimprovementofDOASretrievals

(a)

(b)

Figure6.11:ThefitfactorofthewatereffectasretrievedfromSciamachy(a)andfromGome-2
(b)observations.For(a),thevaluesareaveragedoverthreemonthsfromSeptembertoNovember
2005,andfor(b)thedatafromSeptember2008wereused.Negativefitfactorsdenoteadetection
ofthewatereffectspectrumandthereforeimplythatthelighthaspassedthoughoceanwaterprior
todetection.ItisimportanttonotethesimilarityinthepatternascomparedtoFig.6.10.The
Gome-2datafor(b)havebeenkindlyprovidedbyAndreasRichter,IUPBremen.

172

ConclusionsandrySumma7

Inthepresentstudy,atmosphericcolumnsofiodinemonoxide,IO,havebeenretrievedbyabsorp-
tionspectroscopyonthebasisofradiancespectrarecordedbytheSciamachysensor.Withthe
successfuldevelopmentofthisnewscientifictracegasproduct,theinvestigationofIOamountsand
theirspatialandtemporaldistributionshasbecomepossibleonanearlyglobalscale.Throughthe
workpresentedinthisthesis,progresshasbeenachievedinthefollowingaspects:

EnablingtheretrievalofIOfromSciamachy
TheretrievalofIOfromtheSciamachysensorhasbeendevelopedwithinthisstudyusingofthe
well-establishedDOAS(DifferentialOpticalAbsorptionSpectroscopy)technique.Followingthor-
oughqualityandconsistencyinvestigations,theIOstandardfithasbeendefinedonthebasisofthe
bestresultsfromthesetests.TheidentificationoftheIOabsorptionandthedeterminationofcol-
umnamountsfromsatellitehavethusbecomepossible.Residualsclosetothetheoreticallimitswere
achieved.Anexampleretrieval,whichcoverstheproblematicspectralpositionoftheFraunhofer
G-bandanddoesnotfulfillthequalityandconsistencyrequirements,demonstratesthegeneration
ofsomemisleadingresults.Theinstrument’sdetectionlimitfortheretrievalofIOhasbeendeter-
minedandshowsthatthislimitliesclosetotheIOamountspreviouslyobservedbyground-based
instruments.TheIOabsorptionsignalmaythereforeeasilybehiddenintheinstrumentnoise.The
detectionlimitforasinglemeasurementliesontheorderof7×1012molec/cm2,however,thislimit
isconsiderablyreducedbytemporalaveraging.Seasonalmapsexhibitatypicaldetectionlimitof
1×1012molec/cm2.Forcombinedstatisticalandsystematicinaccuracies,anoverallerrorestimate
of3×1012molec/cm2isdetermined.ThesensitivityofthemeasurementmethodfordetectionofIO
undervaryingconditionshasbeenassessedthroughradiativetransfercalculations.Incomparison
tooceansites,thesensitivitystronglyincreasesforlocationsoverbrightsurfaces,suchasPolarsea
iceregions.Overclearoceanregions,theretrievalofIOfromSciamachyencountersthesame
anomalyassomeotherminortracegasesinthatcolumnamountstendtonegativevalues.Asuit-
ablecorrectionprocedurestillneedstobeestablished.

GlobalobservationsofIOcolumnswithafocusontheAntarctic
ThenewlydevelopedIOproductfromSciamachymeasurementshasenabledtheinvestigationof
theIOglobaldistributionforthefirsttime.Someregionsofspecialinteresthavebeenselectedfor
moredetailedanalysisandamainfocusissetontheAntarcticregion.
TheAntarcticexhibitsthelargestwidespreadIOslantcolumnsglobally.IntheSouthPolarre-
gion,detailedvariationsoftheIOamountsinspaceandtimehavebeenuncovered.ThelargestIO
amountsintheseasonalaveragestypicallyamountto8×1012molec/cm2,whilesomesinglemea-
surementsreachupto2×1013molec/cm2.EnhancedIOisfoundalongtheAntarcticcoast,above

173

ConclusionsandSummary

iceshelvesandregionscoveredbyseaice,aswellasoverpartsofthecontinent.Largestamounts
appearmostlyinspringtime,buttheseasonalcyclevarieswithlocation.Somesimilaritiesare
presentbetweentheIOresultsandbromineoxidedistributions.BrOalsoshowshighestamounts
forAntarcticspringtime,howeverthespatialandtemporaldetailsrevealconsiderabledifferences.
AboveseaiceregionsaroundtheAntarcticcontinent,whereBrOisenhancedbetweenAugustand
October,IOisfoundmainlylaterintheyeararoundNovember.Asecondincreaseinspatialextent
ofenhancedIOamountsaroundMarchisalsonotreflectedintheBrOcolumns.Therevealed
differencesinthespatialandtemporalevolutionsofthetwotracegasesstronglyargueforseparate
releasepathways.WhileaninorganicmechanismisprobablydominatingforBrO,newevidenceis
addedhereforthesuggestionofmainlybiologicalsourcesforIOprecursors.Seaiceconcentration
isreducedinlaterspringascomparedtoearlyspringtime,sothatemissionsofbiogenichalogen
compoundsarefacilitatedbyphytoplanktonsituatedbelowtheseaice.
FromthesatelliteobservationsitbecomesclearthattheNorthernandSouthernHemispheresex-
hibitaconsiderabledifferenceinIOdistributions.NowidespreadsignificantIOamountsare
detectedintheArcticRegion,especiallynotinArcticspring.Thisconstitutesafurtherdifferences
betweenatmosphericIOandBrO,asBrOamountsareequallylargeandwidespreadovertheNorth
PolarspringasintheSouth.AlongsomeNorthernHemisphericcoastlines,enhancedIOcolumns
around4×1012molec/cm2areidentifiedinstronglyconfinedareas.Alignmentofthespecificcoast
lineswiththesatellitegroundpixelspossiblyenhancesthedetectionsensitivityattheselocations
ascomparedtoothercoastswherethegeometryislessfavourableandpotentialIOmightbecon-
cealed.ThelargedifferenceinIOdistributionbetweenthetwohemispheresissurprisingandsofar
notfullyexplained,butdifferentbiologicalconditionsonthetwohemispheresmaybeanimportant
ect.aspEnhancedIOhasbeendetectedabovetheEasternPacific,wherenoIOmeasurementshadbeen
reportedbefore.Concentrationsofdiatomsareincreasedinthisregionwhichstandsinconnection
totheupwellingoceancurrentthere.Diatomsproduceandemitiodocarbons,andprovideapossi-
bleexplanationfortheformationofIOinthatarea.EnhancedIOisnotobservedinotherstrong
upwellingregions,wherediatomconcentrationsaresmaller,sotheconnectiontospecifictypesof
crucial.probablyisytoplanktonph

Resultsfromcomparisonandvalidationstudies
ComparisonswithindependentmeasurementdatasetsofIOhavebeenconductedforvalidation
purposes.SomesuccessfulvalidationstudiesaddconfidencetothenewlydevelopedIOproduct.
Ground-basedLP-DOASobservationsinHalleyResearchStation,Antarctica,revealaseasonalevo-
lutionwhichisingoodagreementwiththeIOresultsfromsatelliteextractedabovethisstation.The
surfacemixingratiosderivedfromthetwodifferentdatasetsagreewellespeciallywhenconsidering
theinvolveduncertainties.Asimilarconclusionisdrawnfromacomparisonwithground-based
datafromaBremenMAX-DOASinstrumentatNy-Ålesund,Spitsbergen,whichispromisingasan
IOconcentrationinthesameorderofmagnitudeofisretrievedfromthegroundandfromsatellite.
TheonlyindependentstudyofIOfromsatellitehasanalysedSciamachymeasurementsforfour
individualdaysabovetheSouthernHemisphere.ThereportedIOcolumnsareconsiderablydiffer-

174

ConclusionsandSummary

entfromtheresultsretrievedinthepresentwork.Usingaretrievalversiontestedwithinthisthesis,
theirresultscouldbereproducedwithsmalleramountsbutwiththesamespatialpattern,which
inthepresentcasecoincideswithlowfitquality.Asaconsequence,thedifferencesarerespected,
butshouldnotresultinalossofconfidenceintheoutcomeofthepresentstudy.
ThedependencyofIOamountsonhighandlowtideatmid-latitudecoastshasbeeninvestigated
fromsatelliteforspecificmid-latitudecoastallocations.However,neitherenhancedIOamountsnor
thetidalsignalisapparentinthesatelliteresults,mostprobablyduetolowIOamountscompared
totheinstrumentaldetectionlimit,stronglylocalisedsourcesandunfavourableobservationalcon-
ditions.

Resultsfromatmosphericchemistrymodellingstudies
TheCAABA/MECCAchemicalboxmodelhasbeenappliedforAntarcticcasestudies.Forvary-
ingprecursorfluxes,theIOsurfaceconcentrationshavebeendetermined.Usingreportedemission
measurements,IOamountsonthesameorderofmagnitudeasobservedfromsatellitecanbere-
produced,butremainratheronthelowsideevenforthehighestreportedprecursorfluxes.No
precursorfluxesareavailableforthoseregionswithlargestsatelliteIOcolumnsthough.Inagree-
mentwithexpectations,theemissionsofpoly-halogenatedcarbonsarecrucialfortheproduction
ofsufficientIOamounts.Infuturestudies,someadditionalchemicalreactionpathwayswillbe
includedintothechemicalmechanism,e.g.,somerecyclingofgaseousiodinefromtheparticulate
phase,andalsothephotolysisofOIO.

studiesfutureforectsProsp•ThetimeseriesofthepresentstandardIOproductwillbefurtherextendedforthestillongoing
measurementsofSciamachy,increasingtheIOdatabase.Furthervalidationandcomparison
studiesusingindependentdatasetsshallcoveradditionallocations,e.g.,fortheEasternPacific
whererecentcampaignsrevealenhancedIOamountsfromship-borneinstrumentation.
•FurtherimprovementsonthesatellitedatafromSciamachymaybecomepossibleifthefitting
windowmaybeextendedbeyondthe431nmregiontoincludemoreabsorptionbandsoftheIO
spectrum.Forthat,betterunderstandingandcompensationofthespectralcharacteristicsof
atmosphericradiationandoftheinstrument’sresponseandpolarisationsensitivityisrequired.
•Datafromadditionalsatelliteinstrumentsmaybehelpfulinaddingevenmoreinformation
abouttheatmosphericIOcontent.Especially,IOretrievalsfromGOME-2wouldbean
interestingproductasthespatialcoverageisbetterthanforSciamachy.Itisnotknownyet
ifthesignal-to-noisequalityofGOME-2willsufficefortheretrievalofsuchaminortracegas
asIO.Foradeeperinsightintocertainregionsandevendailyvariations,aspectrometerona
geostationarysatellitewouldmakeanimprovementasthedataamountpergroundlocation
isstronglyenhanced.Ofcoursethiswillnotyieldglobalcoverage,andisaprospectforthe
future.tdistanmore•InordertobetterquantifytheobservedIOcolumns,especiallyintermsofsurfaceconcentra-
tions,detailedinformationontheIOprofileisneeded.Thisinformationenterstheairmass

175

ConclusionsandSummary

176

factorcalculationaswellasthetransformationfromaverticalcolumnintoamixingratio.
Currently,however,theuncertaintyontheIOaltitudeprofileislarge,sothatfurtherfield
crucial.aretsmeasuremen

•ForimprovedinterpretationpossibilitiesoftheIOobservations,alargerbasisofprecursor
fluxmeasurementsinsourceregionsisneededaswellaslocalmeasurementsofIOinspecific
regions,e.g.,withintheAntarcticseaicezoneorontheAntarcticcontinent,inbiologically
activeoceanareasandatcoastalsites,preferablyoveralongerperiodoftimethanthetypical
campaignbasistoalsoassesstheseasonalbehaviour.

•Someinvestigatedaspectsinthepresentstudyhaveremainedbelowthedetectionlimitof
theSciamachysensorpartlyduetothecomparativelylargegroundpixel.Ontheother
hand,localgroundbasedmeasurementsarerestrictedinspatialcoverage.Thisgapmay
beclosedinthefuturebyapplyingairbornemeasurements.Forspecificresearchtasks,the
constructionofanewinstrumentiscurrentlyongoing,whichisscheduledforobservations
fromanaircraftplatform.TheplanningandopticaltestingofaDOASsystembasedon
animagingspectrometerisproceeding,andthepossibilityofacustom-builtspectrometer
usingaholographicreflectiongratingisassessed.Theconstructionwithalargeopeningangle
(smallF-number)andareducednumberofopticalelementsisexpectedtoyieldgoodspatial
coverageandstronglightthroughput.Withthisinstrumentandthemeasurementconcept
ofasmallandlowflyingaircraft,afinegroundspatialresolutionwillbeachieved,which
providespromisingconditionsforthemeasurementoflocalisedIOabundances.Moredetailed
observationsofthespatialIOdistribution,e.g.,alongcoastlinesinthemid-latitudesaswell
asthedetectionoflocalisedIOsourceregionsandsourcestrengthswillbecomepossiblewith
t.instrumenthis

reviationsabbofListAMSR-EAdvancedMicrowaveScanningRadiometerforEOS
ASMAzimuthScanMechanisms/Mirror
CAABAChemistryAsABoxmodelApplication
onsChloro-Fluoro-CarbCCFDOASDifferentialOpticalAbsorptionSpectroscopy
SystemObservingEarthEOSAgencySpaceeanEuropESAESMElevationScanMechanisms/Mirror
ESMEarthSystemModel(Chapter5)
delMoCirculationGlobalGCMGOME-2GlobalOzoneMonitoringInstrument-2
HITRANHIgh-resolutionTRANsmissionmolecularabsorptiondatabase
IVOCIodinatedVolatileOrganicCompound
cessorPre-ProKineticKPPtSighOfLineLOSLP-DOASLong-PathDoas
asDoMulti-AXisASMAX-DOMECCAModuleEfficientlyCalculatingtheChemistryoftheAtmosphere
MBLMarineBoundaryLayer
NASANationalAeronauticsandSpaceAdministration
Real-TimeNearTNRODEOzoneDepletionEvent
ODEOrdinaryDifferentialEquation(Chapter5)
tInstrumenMonitoringOzoneOMITimeosureExpPixelPETScatteringRamanRotationalRRSSciamachySCanningImagingAbsorptionspectroMeterforAtmosphericCHartographY
SensorField-Of-viewWIdeSea-viewingSeaWIFSSPICSSciamachy-PMDbasedIdentificationandclassificationofCloudsandSurfaces
AngleZenithSolarSZAVHOCVolatileHalogenatedOrganicCompound
VOCVolatileOrganicCompound
VOIVolatileOrganicIodine
ScatteringRamanVibrationalVRSVSLSVeryShort-LivedSpecies

177

Bibliography

Alicke,B.,Hebestreit,K.,Stutz,J.,andPlatt,U.:Iodineoxideinthemarineboundarylayer,Nature,397,
1999.572–573,Allan,B.J.,McFiggans,G.,Plane,J.M.C.,andCoe,H.:Observationsofiodinemonoxideintheremote
marineboundarylayer,J.Geophys.Res.,105,14363–14369,2000.
Allan,B.J.,Plane,J.M.C.,andMcFiggans,G.:ObservationsofOIOintheremotemarineboundarylayer,
Geophys.Res.Lett.,28,1945–1948,2001.
Alvain,S.,Moulin,C.,Dandonneau,Y.,andBréon,F.:Remotesensingofphytoplanktongroupsincase1
watersfromglobalseawifsimagery,DeepSeaResearch,I52,1989–2004,2005.
Amachi,S.,Kamagata,Y.,Kanagawa,T.,andMaramatsu,Y.:Bacteriamediatemethylationofiodinein
marineandterrestrialenvironments,Appl.Env.Microbiol.,67,2718–2722,2001.
Andreae,M.O.,Atlas,E.,Harris,G.W.,Helas,G.,deKock,A.,Koppmann,R.,Maenhaut,W.,Manø,
S.,Pollock,W.H.,Rudolph,J.,Scharffe,D.,Schebeske,G.,andM.,W.:Methylhalideemissionsfrom
savannafiresinsouthernafrica,JournalofGeophysicalResearch,101,23603–23613,1996.
AM.tkinson,J.,andR.,TroBaulce,h,J.:D.EvL.,aluatedCox,R.kineticA.,Croandwleyphoto,J.chN.,emicalHampson,dataforR.F.,atmosphericHynes,R.cG.,hemistry:Jenkin,VoM.lumeE.,I-Rossi,gas
phasereactionsofOx,HOx,NOxandSOxspecies,AtmosphericChemistryandPhysics,4,1461–1738,
2004.Atkinson,R.,Baulch,D.L.,Cox,R.A.,Crowley,J.N.,Hampson,R.F.,Hynes,R.G.,Jenkin,M.E.,
Rossi,M.J.,Troe,J.,andSubcommittee,I.:Evaluatedkineticandphotochemicaldataforatmospheric
chemistry:VolumeII-gasphasereactionsoforganicspecies,AtmosphericChemistryandPhysics,6,
2006.3625–4055,Barrie,L.A.,Bottenheim,J.W.,Schnell,R.C.,Crutzen,P.J.,andRasmussen,R.A.:Ozonedestruction
andphotochemicalreactionsatpolarsunriseinthelowerArcticatmosphere,Nature,334,138–141,1988.
Bates,T.S.,Lamb,B.K.,Guenther,A.,Dignon,J.,andStoiber,R.E.:Sulfuremissionstotheatmosphere
fromnaturalsources,JournalofAtmosphericChemistry,14,315–337,1992.
Begoin,M.,Richter,A.,Kaleschke,L.,Tian-Kunze,X.,Stohl,A.,andBurrows,J.P.:Satelliteobservations
oflongrangetransportofalargeBrOcloudintheArctic,AtmosphericChemistryandPhysics,submitted
2009.August,Bloss,W.J.,Rowley,D.M.,Cox,R.A.,andJones,R.L.:KineticsandproductsoftheIOself-reaction,J.
Phys.Chem.A,105,7840–7854,2001.
Bloss,W.J.,Lee,J.D.,Johnson,G.P.,Sommariva,R.,Heard,D.E.,Saiz-Lopez,A.,McFiggans,G.,Coe,
H.,Flynn,M.,Williams,P.,Rickard,A.R.,andFleming,Z.L.:Impactofhalogenmonoxidechemistry
uponboundarylayerOHandHO2concentrationsatacoastalsite,GeophysicalResearchLetters,32,
2005.doi:10.1029/2004GL022084,814,L06Bobrowski,N.,Hönninger,G.,Galle,B.,andPlatt,U.:Detectionofbrominemonoxideinavolcanicplume,
2003.doi:10.1038/nature01625,273–276,423,Nature,Bogumil,K.,Orphal,J.,Homann,T.,Voigt,S.,Spietz,P.,Fleischmann,O.C.,Vogel,A.,Hartmann,M.,
Bovensmann,H.,Frerik,J.,andBurrows,J.P.:Measurementsofmolecularabsorptionspectrawith
thesciamachypre-flightmodel:Instrumentcharacterizationandreferencedataforatmosphericremote-
sensinginthe230-2380nmregion,J.Photochem.Photobiol.A,157,2003.

178

Bovensmann,H.,Burrows,J.P.,Buchwitz,M.,Frerick,J.,Noël,S.,Rozanov,V.V.,Chance,K.V.,and
Goede,A.P.H.:Sciamachy:Missionobjectivesandmeasurementmodes,J.Atmos.Sci.,56,127–150,
1999.Bracher,A.,Vountas,M.,Dinter,T.,Burrows,J.P.,Röttgers,R.,andPeeken,I.:Quantitativeobservation
ofcyanobacteriaanddiatomsfromspaceusingphytodoasonsciamachydata,Biogeosciences,6,751–764,
2009.Brasseur,G.P.,Orlando,J.J.,andTyndall,G.S.:AtmososphericChemistryandGlobalChange,Oxford
1999.Press,yersitUnivBrewer,A.W.,McElroy,C.T.,andKerr,J.B.:Nitrogendioxideconcentrationsintheatmosphere,Nature,
1973.129–133,246,Brinkmann,R.:Rotationalramanscatteringinplanetaryatmospheres,AstrophysicalJournal,154,1087–
1968.1093,Bösch,H.,Camy-Peyret,C.,Chipperfield,M.P.,Fitzenberger,R.,Harder,H.,Platt,U.,andPfeilsticker,
bK.:orneUppsolareroclimitsofcultationstratosphericvisiblespIOectra:andOIOImplicationsinferredforfromtotalcengaseouster-to-limiodineb-darkandening-correctedstratosphericozone,balloon-J.
Geophys.Res.,108,doi:10.1029/2002JD003078,2003.
Burrows,J.P.,Hölzle,E.,Goede,A.P.H.,Visser,H.,,andFricke,W.:Sciamachy-scanningimaging
absorptionspectrometerforatmosphericchartography,ActaAstronautica,35,445–451,1995.
Burrows,J.P.,Dehn,A.,Deters,B.,Himmelmann,S.,Richter,A.,Voigt,S.,andOrphal,J.:Atmospheric
remotesensingreferencedatafromgome:Part1temperaturedependentabsorptioncross-sectionsofno2
inthe231-794nmrange,JournalofQuantitativeSpectroscopyandRadiativeTransfer,60,1025–1031,
1998.Burrows,J.P.,Richter,A.,Dehn,A.,Deters,B.,Himmelmann,S.,Voigt,S.,andOrphal,J.:Atmospheric
remote-sensingreferencedatafromgome:Part2.temperature-dependentabsorptioncrosssectionsofo3
inthe231-794nmrange,JournalofQuantitativeSpectroscopyandRadiativeTransfer,61,509–517,
1999a.doi:10.1016/S0022-4073(98)00037-5,Burrows,J.P.,Weber,M.,Buchwitz,M.,Rozanov,V.V.,Ladstätter-Weißenmayer,A.,Richter,A.,DeBeek,
R.,Hoogen,R.,Bramstedt,K.,andEichmann,K.U.:Theglobalozonemonitoringexperiment(gome):
Missionconceptandfirstscientificresults,J.Atmos.Sc.,56,151–175,1999b.
Butz,C.,A.,SchwBöscärzle,h,H.,J.,Camandy-PPfeilsticeyret,kC.,er,K.:Chipperfield,ConstrainM.tsP.on,Dorf,inorganicM.,Kreycygaseous,S.,iodineKritten,inL.,thetropicalPrados-Román,upper
troposphereandstratosphereinferredfromballoon-bornesolaroccultationobservations,Atmospheric
ChemistryandPhysicsDiscussions,9,14645–14681,2009.
Carpenter,L.:(aseditor)Interactivecommenton”Amechanismforbiologically-inducediodineemissions
fromsea-ice”byA.Saiz-LopezandC.S.Boxe,AtmosphericChemistryandPhysics,8,S1316–S1316,
2008.Carpenter,L.J.,Sturges,W.T.,Penkett,S.A.,Liss,P.S.,Alicke,B.,Hebestreit,K.,andPlatt,U.:Short-
livedalkyliodidesandbromidesatMaceHead,Ireland:Linkstobiogenicsourcesandhalogenoxide
production,J.Geophys.Res.,104,1679–1689,1999.
Carpenter,L.J.,Hebestreit,K.,Platt,U.,,andLiss,P.S.:CoastalzoneproductionofIOprecursors:a
2-dimensionalstudy,Atmos.Chem.Phys.,1,9–18,2001.
Carpenter,L.J.,Hopkins,J.R.,Jones,C.E.,Lewis,A.C.,Parthipan,R.,Wevill,D.J.,Poissant,L.,Pilote,
M.,andConstant,P.:Abioticsourceofreactiveorganichalogensinthesub-arcticatmosphere?,Environ.
Sci.Technol.,39,8812–8816,2005.
Carpenter,L.J.,Wevill,D.J.,Palmer,C.J.,andMichels,J.:Depthprofilesofvolatileiodineandbromine-
containinghalocarbonsincoastalantarcticwaters,MarineChemistry,103,227–236,2007.

179

Chamberlain,A.C.:Depositionofiodine-131innorthernenglandinoctober1957,Quart.J.Roy.Met.Soc.,
1960.350–361,85,Chamberlain,A.C.,Eggleton,A.E.J.,Megaw,W.J.,andMorris,J.B.:Behaviourofiodinevapourinair,
Disc.FaradaySoc.,30,162–169,1960.
Chameides,W.L.andDavis,D.D.:Iodine:itspossibleroleintroposphericchemistry,J.Geophys.Res.,
1980.7383–7398,85,Chance,K.:AnalysisofBrOmeasurementsfromtheGlobalOzoneMonitoringExperiment,Geophys.Res.
1998.3335–3338,25,Lett.,Chapman,S.:Atheoryoftheupper-atmosphereozone,Mem.Roy.Meteorol.Soc.,3,103,1930.
Charlson,R.J.,Lovelock,J.E.,Andreae,M.O.,andWarren,S.G.:Oceanicphytoplankton,atmospheric
sulfur,cloudalbedoandclimate,Nature,326,655–661,1987.
Chatfield,R.B.andCrutzen,P.J.:Arethereinteractionsofiodineandsulfurspeciesinmarineairphoto-
chemistry?,J.Geophys.Res.,95,22319–22341,1990.
Chubachi,S.:PreliminaryresultofozoneobservationsatSyowastationfromFebruary1982toJanuary
1983,Mem.Natl.Inst.PolarRes.[Spec.IssueJpn.34],pp.13–19,1984.
Cox,R.A.andCoker,G.B.:AbsorptioncrosssectionandkineticsofiointhephotolysisofCH3Iinthe
presenceofozone,J.Phys.Chem.,87,4478–4484,1983.
Cox,R.A.,Bloss,W.J.,Jones,R.L.,andRowley,D.M.:OIOandtheatmosphericcycleofiodine,Geophys.
1999.1857–1860,26,Lett.,Res.Damian,V.,Sandu,A.,Damian,M.,Potra,F.,andCarmichael,G.:Thekineticpreprocessorkpp–a
softwareenvironmentforsolvingchemicalkinetics,ComputersandChemicalEngineering,26,1567–1579,
2002.Datenvis,tialD.,Craimpactwofford,ioJ.,dineLiu,onS.,tropMcKeen,osphericS.,levelsBandyof,ozoneA.,andThornotherton,D.,criticalRoowland,xidants,F.S.,JournalandofBlake,GeophD.:ysicalPo-
1996.2135–2147,101D,h,ResearcdeBeek,R.,Vountas,M.,Rozanov,V.,Richter,A.,andBurrows,J.:Theringeffectinthecloudyatmo-
sphere,Geophys.Res.Lett.,28(4),721–724,2001.
Demtröder,W.:ExperimentalphysikBd.3-Atome,MoleküleundFestkörper,Springer,2ndedition,2000.
DickOzoneerson,inR.theR.,remoteRhoads,marineK.P.b,oundaryCarsey,laT.yer:P.,apOltmans,ossibleS.roleJ.,forCrutzen,halogens,P.J.,J.andGeophBurroys.ws,Res,J.104,P.:
1999.doi:10.1029/1999JD900023,Duce,R.,Wasson,J.T.,Winchester,J.W.,andBurns,F.:Atmosphericiodine,bromineandchlorine,J.
Geophys.Res,68,3943–3947,1963.
EuropeanParliament:Directive2002/3/ECoftheEuropeanParliamentandoftheCouncilof12February
2002relatingtoozoneinambientair,OfficialJournaloftheEuropeanCommunities,2002.
Farman,J.C.,Gardiner,B.G.,andShanklin,J.D.:LargelossesoftotalozoneinAntarcticarevealseasonal
ClOx/NOxinteraction,Nature,315,207–210,1985.
Fleischmann,O.C.,Hartmann,M.,Burrows,J.P.,andOrphal,J.:Newultravioletabsorptioncross-sections
ofBrOatatmospherictemperaturesmeasuredbyatime-windowingfouriertransformspectroscopy,J.
Photochem.Photobiol.A.,168,117–132,2004.
Francis,P.,Burton,M.R.,andOppenheimer,C.:Remotemeasurementsofvolcanicgascompositionsby
solaroccultationspectroscopy,Nature,396,567–270,doi:10.1038/25115,1998.

180

Friess,U.,Wagner,T.,Pundt,I.,Pfeilsticker,K.,andPlatt,U.:Spectroscopicmeasurementsoftropospheric
iodineoxideatneumeyerstation,antarctica,Geophys.Res.Lett.,28,doi:10.1029/2003JD004133,2001.
Giese,B.,L.F.A.F.C.andWiencke,C.:Releaseofvolatileiodinatedc1-c4hydrocarbonsbymarine
macroalgaefromvariousclimatezones,EnvironmentalScienceandTechnology,33,2432–2439,1999.
GómezMartín,J.C.,Spietz,P.,andBurrows,J.P.:Spectroscopicstudiesofthei2/o3photochemistrypart
1:Determinationoftheabsoluteabsorptioncrosssectionsofiodineoxidesofatmosphericrelevance,J.
Photochem.Photobiol.A,176,15–38,doi:10.1016/j.jphotochem.2005.09.024,2005.
GómezMartín,J.C.,Spietz,P.,andBurrows,J.P.:Kineticandmechanisticstudiesofthei2/o3photo-
chemistry,J.Phys.Chem.A.,111,doi:10.1021/jp061186c,2007.
GómezMartín,J.C.,Ashworth,S.H.,Mahajan,A.S.,andPlane,J.C.M.:PhotochemistryofOIO:Labo-
ratorystudyandatmosphericimplications,Geophys.Res.Lett.,36,L09802,doi:10.1029/2009GL037642,
2009.Gottwald,M.,Bovensmann,H.,Lichtenberg,G.,Noel,S.,vonBargen,A.,Slijkhuis,S.,Piters,A.,
Hoogeveen,R.,vonSavigny,C.,Buchwitz,M.,Kokhanovsky,A.,Richter,A.,Rozanov,A.,Holzer-Popp,
T.,Bramstedt,K.,Lambert,J.-C.,Skupin,J.,Wittrock,F.,Schrijver,H.,andBurrows,J.P.,eds.:SCIA-
MACHY,MonitoringthechangingEarth’satmosphere,DLR,InstitutfürMethodikderFernerkundung
2006.(IMF),Grainger,J.F.andRing,J.:Anomalousfraunhoferlineprofiles,Nature,193,762,1962.
oGreenxygenblatt,betG.weenD.,330Orlando,andJ.1140J.,nm,J.Burkholder,GeophJ.ys.B.,Res.,andRa95(D11),vishank18ara,577–18A.R.:582,1990.Absorptionmeasurementsof
Hecht,E.:Optik,OldenbourgVerlag,3edn.,2001.
Heckel,A.,Richter,A.,Tarsu,T.,Wittrock,F.,Hak,C.,Pundt,I.,Junkermann,W.,andBurrows,J.P.:
Max-doasmeasurementsofformaldehydeinthepo-valley,AtmosphericChemistryandPhysics,5,909–
2005.918,Hill,V.L.andManley,S.L.:Releaseofreactivebromineandiodinefromdiatomsanditspossiblerolein
halogentransferinpolarandtropicaloceans,Limnol.Oceanogr.,54,812–822,2009.
Hönninger,G.,Leser,H.,Sebastián,O.,andPlatt,U.:Ground-basedmeasurementsofhalogenox-
idesattheHudsonBaybyactivelongpathDOASandpassiveMAX-DOAS,Geophys.Res.Lett,31,
2004.doi:10.1029/2003GL018982,Hooker,S.,Esaias,W.,Feldman,G.,Gregg,W.,andMcClain,C.:AnOverviewofSeaWiFSandOcean
Color,NASATech.Memo.104566,Vol.1,S.B.HookerandE.R.Firestone,Eds.,NASAGoddardSpace
FlightCenter,Greenbelt,Maryland,24pp.,pluscolorplates,1992.
IPCC:ClimateChange2007:ThePhysicalScienceBasis.ContributionofWorkingGroupItotheFourth
Manning,AssessmentM.,RepandortofChen,theZ.,InandtergovMarquis,ernmenM.,talPandanelAvoneryt,K.ClimateB.,andChange.Tignor,Solomon,M.,andS.,Miller,andH.Qin,L.D.,(eds.),and
CambridgeUniversityPress,2007.
Jöckel,P.,Sander,R.,Kerkweg,A.,Tost,H.,andLelieveld,J.:Technicalnote:Themodularearthsubmodel
system(messy)-anewapproachtowardsearthsystemmodeling,Atmos.Chem.Phys.,5,433–444,2005.
Jimenez,J.L.,Bahreini,R.,CockerIII,D.R.,Zhuang,H.,Varutbangkul,V.,Flagan,R.C.,Seinfeld,
J.H.,O’Dowd,C.D.,andHoffmann,T.:Newparticleformationfromphotooxidationofdiiodomethane
(CH2I2),J.Geophys.Res.,108,4318,doi:10.1029/2002JD002452,2003.
Johnston,P.V.andMcKenzie,R.L.:NO2observationsat45◦Sduringthedecreasingphaseofsolarcycle
21,from1980to1987,JournalGeophysicalResearch,94,3473–3486,1989.
Joiner,satelliteJ.andbackscatterBhartia,Pultra.K.:violetThemeasuremendeterminationts,J.ofGeophcloudys.Res.,pressures100,from23019–23rotational026,1995.ramanscatteringin

181

Kaleschke,L.,Richter,A.,Burrows,J.P.,Afe,O.,Heygster,G.,Notholt,J.,Rankin,A.M.,Roscoe,H.K.,
Hollwedel,J.,Wagner,T.,andJacobi,H.-W.:Frostflowersonseaiceasasourceofseasaltandtheir
influenceontropospherichalogenchemistry,Geophys.Res.Lett.,31,doi:10.1029/2004GL020655,2004.
Küpper,F.,Schweigert,N.,ArGall,E.,Legendre,J.-M.,Vilter,H.,andKloareg,B.:Iodineuptakein
laminarialesinvolvesextracellular,haloperoxidase-mediatedoxidationofiodide,Planta,207,163–171,
1998.Küpper,F.C.,Carpenter,L.J.,McFiggans,G.B.,Palmer,C.J.,Waite,T.J.,Boneberg,E.-M.,Woitsch,
S.,Weiller,M.,Abela,R.,Grolimund,D.,Potin,P.,Butler,A.,III,G.W.L.,Kroneck,P.M.H.,
Meyer-Klaucke,W.,andFeiters,M.C.:Iodideaccumulationprovideskelpwithaninorganicantioxidant
impactingatmosphericchemistry,ProceedingsoftheNationalAcademyofSciences,105,6954–6958,2008.
Krijger,J.M.,Aben,I.,andSchrijver,H.:Distinctionbetweencloudsandice/snowcoveredsurfacesinthe
identificationofcloud-freeobservationsusingsciamachypmds,AtmosphericChemistryandPhysics,5,
2005.2729–2738,Kurucz,R.,Furenlid,I.,Brault,J.,andTesterman.,L.:SolarFluxAtlasfrom296to1300nm,National
SolarObservatory,1984.
Levelt,P.F.,Hilsenrath,E.,Leppelmeier,G.W.,vandenOord,G.H.J.,Bhartia,P.K.,Tamminen,
J.,deHaan,J.F.,andVeefkind,J.P.:Scienceobjectivesoftheozonemonitoringinstrument,IEEE
TransactionsonGeoscienceandRemoteSensing,44,doi:10.1109/TGRS.2006.872336.,2006.
Lillian,D.,BirSingh,H.,Appleby,A.,Lobban,L.,Arnts,R.,Gumpert,R.,Hague,R.,Toomey,J.,Kazazis,
J.,Antell,M.,Hansen,D.,andScott,B.:Atmosphericfatesofhalogenatedcompounds,Environ.Sci.
Tech.,9,1042–1048,1975.
Lotz,W.A.,Vountas,M.,Dinter,T.,andBurrows,J.P.:Cloudandsurfaceclassificationusingsciamachy
polarizationmeasurementdevices,Atmos.Chem.Phys.Discuss.,8,9855–9881,2008.
Lovelock,J.E.:HandsupfortheGaiahypothesis,Nature,344,100–102,1990.
Lovelock,J.E.,Maggs,R.J.,andWade,R.J.:Halogenatedhydrocarbonsinandovertheatlantic,Nature,
1973.194–196,241,Mahaney,W.C.,Dohm,J.M.,Baker,V.R.,Newsom,H.E.,Malloch,D.,Hancock,R.,Campbell,I.,
Sheppard,D.,andMilner,M.W.:Antarcticmicrobescolonizeundermars-likeconditions,Tech.rep.,
2002.df,dffiles/news_233.pttp://www.astrobio.net/phMartino,M.,Mills,G.P.,Woeltjen,J.,andLiss,P.S.:Anewsourceofvolatileorganoiodinecompoundsin
surfaceseawater,GeophysicalResearchLetters,36,doi:10.1029/2008GL036334,2009.
ofMcFiggans,iodinecG.,hemistryPlane,inJ.theM.C.,marineAllan,bB.oundaryJ.,laCarpyer,enJ.ter,L.GeophJ.,ys.Coe,Res.,H.,105,and14O’Do371–14wd,C.:385,Am2000.odelingstudy
McFiggans,G.,Coe,H.,Burgess,R.,Allan,J.,Cubison,M.,Alfarra,M.R.,Saunders,R.,Saiz-Lopez,
A.,Plane,J.M.C.,Wevill,D.J.,Carpenter,L.J.,Rickard,A.R.,andMonks,P.S.:Directevidence
forcoastaliodineparticlesfromlaminariamacroalgae-linkagetoemissionsofmoleculariodine,Atmos.
Chem.Phys.,4,701–713,2004.
Mäkpinen,elä,J.E.M.,I.,andHoffmann,O’Dowd,T.,C.HolzkD.:e,C.,BiogenicVäkevioä,dineM.,Suni,emissionsT.,andMattila,idenT.,tificationAalto,ofP.P.end-pro,Tappductser,inU.,coastalKaup-
ultrafineparticlesduringnucleationbursts,J.Geophys.Res.,107,8110,doi:10.1029/2001JD000580,2002.
Molina,M.J.andRowland,F.S.:Stratosphericsinkforchlorofluoromethanes:chlorineatomcatalysed
destructionofozone,Nature,249,810–814,1974.
Moore,R.M.,Webb,M.,Tokarczyk,R.,andWever,R.:Bromoperoxidaseandiodoperoxidaseenzymes
andproductionofhalogenatedmethanesinmarinediatomcultures,J.Geophys.Res.,101,20899–20908,
1996.

182

Murphy,D.M.,Thomson,D.S.,andMiddlebrook,A.M.:Bromine,iodine,andchlorineinsingleaerosol
particlesatcapegrim,Geophys.Res.Lett.,24,3197–3200,1997.
Noël,S.,Bovensmann,H.,Burrows,J.P.,Frerick,J.,Chance,K.V.,Goede,A.H.P.,andMuller,C.:The
sciamachyinstrumentonenvisat-1,sensors,systems,andnext-generationsatellitesii,Proc.SPIE,3498,
1998.94–104,Noxon,J.F.:Nitrogendioxideinthestratosphereandtropospheremeasuredbyground-basedabsorption
spectroscopy,Science,189,547,1975.
Noxon,J.F.,Whipple,Jr.,E.C.,andHyde,R.S.:StratosphericNO2,1.observationalmethodandbehavior
atmidlatitude,J.Geophys.Res.,84,5047–5065,1979.
O’Dowd,C.D.andHoffmann,T.:Coastalnewparticleformation:Areviewofthecurrentstate-of-the-art,
Environ.Chem.,2,doi:10.1071/EN05077,2005.
O’Dowd,C.D.,McFiggans,G.,Creasey,D.J.,Pirjola,L.,Hoell,C.,Smith,M.H.,Allan,B.J.,Plane,J.
M.C.,Heard,D.E.,Lee,J.D.,Pilling,M.J.,andKulmala,M.:Onthephotochemicalproductionof
newparticlesinthecoastalboundarylayer,Geophys.Res.Lett.,26,1707–1710,1999.
O’Dowd,C.D.,Hämeri,K.,Mäkelä,J.M.,Pirjola,L.,Kulmala,M.,Jennings,S.G.,Berresheim,H.,Hans-
son,mann,H.-C.,T.:AdeLeeudedicatedw,G.,studyKunz,ofnewG.J.,particleAllen,A.formationG.,andHewitt,fateC.inN.,theJaccoastalkson,enA.,vironmenViisanen,t(PY.,ARFandORCE):Hoff-
Overviewofobjectivesandachievements,J.Geophys.Res.,107,8108,doi:10.1029/2001JD000555,2002a.
O’Dowd,C.D.,Jimenez,J.L.,Bahreini,R.,Flagan,R.C.,Seinfeld,J.H.,Hameri,K.,Pirjola,L.,Kulmala,
M.,Jennings,S.G.,andHoffmann,T.:Marineaerosolformationfrombiogeniciodineemissions,Nature,
2002b.632–636,417,Oetjen,H.:Measurementsofhalogenoxidesbyscatteredsunlightdifferentialopticalabsorptionspectroscopy,
Ph.D.thesis,UniversityBremen,2009.
Pedersén,M.,Collén,J.,Abrahamsson,K.,andEkdahl,A.:Productionofhalocarbonsbyseaweeds:an
oxidativestressreaction?,ScientiaMarina,60,257–263,1996.
PJ.,eters,andC.,PPlatt,echtl,U.:S.,ReactivStutz,eJ.,andHeborganicestreit,halogenK.,spHönninger,eciesinG.,threeHeumann,differentK.europG.,eanSchwcoastalarz,enA.,Winvironmenterlik,ts,
Atmos.Chem.Phys.,5,3357–3375,2005.
Piot,M.andvonGlasow,R.:Thepotentialimportanceoffrostflowers,recyclingonsnow,andopenleads
forozonedepletionevents,Atmos.Chem.Phys.,8,2437–2467,2008.
Platt,U.andHönninger,G.:Theroleofhalogenspeciesinthetroposphere,Chemosphere,52,325–338,
2003.Platt,U.andPerner,D.:DirectmeasurementsofatmosphericCH2O,HNO2,O3,NO2,SO2bydifferential
opticalabsorptioninthenearUV„JournalGeophysicalResearch,85,7453–7458,1980.
Platt,U.andStutz,J.:DifferentialOpticalAbsorptionSpectroscopy-PrinciplesandApplications,Springer-
2008.erlag,VPlatt,tationU.andplan,vonGeophGlasoys.w,Res.R.:Abstr.,Halogens7,inWhitethePtropaper,osphereh-aninttp://www.igac.noaa.goternationaltask.sciencev/HitT_WhitePandapimplemen-er.pdf,
2005.Pope,R.M.andFry,E.:Absorptionspectrum(380-700nm)ofpurewater.ii.integratingcavitymeasure-
ments,AppliedOptics,36,1997.
Rasmussen,R.A.,Khalil,M.A.K.,Gunawardena,R.,andHoyt,D.:Atmosphericmethyliodide(CH3I),
JournalofGeophysicalResearch,87,2086–3090,1982.

183

Ravishankara,A.R.,Solomon,S.,Turnipseed,A.A.,andWarren,R.F.:Atmosphericlifetimesoflong-lived
halogenatedspecies,Science,259,1993.
Read,K.A.,Mahajan,A.S.,Carpenter,L.J.,Evans,M.J.,Faria,B.V.E.,Heard,D.E.,Hopkins,J.R.,
M.Lee,J.,J.andD.,Plane,Moller,J.S.M.J.,C.:Lewis,ExtensivA.eC.,Mendes,halogen-mediatedL.,McQuaid,ozoneJ.destructionB.,Oetjen,overH.,thetropicalSaiz-Lopez,atlanA.,ticoPilling,cean,
2008.doi:10.1038/nature07035,1232–1235,453,Nature,Reifenhäuser,W.andHeuman,K.G.:Determinationofmethyliodideintheantarcticatmosphereandthe
southpolarsea,Atmos.Environ.,26A,2905–2912,1992.
Richter,A.:AbsorptionsspektrospokischeMessungenstratosphärischerSpurengaseüberBremen,53◦N,
Ph.D.thesis,UniversityBremen,1997.
Richter,A.,Wittrock,F.,Eisinger,M.,andBurrows,J.P.:GOMEobservationsoftroposphericBrOin
NorthernHemisphericspringandsummer1997,Geophys.Res.Lett,25,2683–2686,1998.
Richter,A.,Eyring,V.,Burrows,J.P.,Bovensmann,H.,Lauer,A.,Sierk,B.,andCrutzen,P.J.:
SatellitemeasurementsofNO2frominternationalshippingemissions,Geophys.Res.Lett.,31,
2004.doi:10.1029/2004GL020822,Richter,A.,Burrows,J.P.,Nüß,H.,Granier,C.,andNiemeier,U.:Increaseintroposphericnitrogendioxide
overchinaobservedfromspace,Nature,437,doi:10.1038/nature04092,2005.
Robertson,H.A.andFalconer,I.R.:Accumulationofradioactiveiodineinthyroidglandssubsequentto
nuclearweapontestsandtheaccidentatwindscale,Nature,184,1699–1702,1959.
Rozanov,A.,Bovensmann,H.,Bracher,A.,Hrechanyy,S.,Rozanov,V.,Sinnhuber,M.,Stroh,F.,and
Burrows,J.P.:NO2andBrOverticalprofileretrievalfromSCIAMACHYlimbmeasurements:Sensi-
tivitystudies,AdvancesinSpaceResearch,36,846–854,AtmosphericRemoteSensing:Earth’sSurface,
Troposphere,StratosphereandMesosphere-I,2005a.
Rozanoradiativv,eA.,transferRozanomov,delV.-V.,forBucgeophhwitz,ysicalM.,Kapplicationsokhanoinvsky,theA.,and175-2400Burronmws,spJ.ectralP.:Sregion,CIAATRANdv.2.0.SpaceARes.,new
2005b.doi:10.1016/j.asr.2005.03.012,1015–1019,36,Rozanov,V.V.,Diebel,D.,Spurr,R.J.D.,andBurrows,J.P.:Gometran:Aradiativetransfermodelfor
thesatelliteprojectGOME,theplane-parallelversion,J.Geophys.Res.,102,16683–16695,1997.
Saiz-Lopez,A.andPlane,J.M.C.:Noveliodinechemistryinthemarineboundarylayer,Geophysical
RsearchLetters,31,L04112,doi:10.1029/2003GL019215,2004.
andSaiz-Lopno3ez,inA.,theShillito,mid-latitudeJ.A.,Coe,marineH.,bandoundaryPlane,laJ.yer,M.AC.:tmos.Chem.MeasuremenPhtsys.,and6,mo1513–1528,dellingofi2006.2,io,oio,bro
Saiz-Lopez,A.,Chance,K.,Liu,X.,Kurosu,T.P.,andSander,S.P.:Firstobservationsofiodineoxide
fromspace,Geophys.Res.Lett.,34,doi:doi:10.1029/2007GL030111,2007a.
Saiz-Lopez,A.,Mahajan,A.S.,Salmon,R.A.,Bauguitte,S.J.-B.,,Jones,A.E.,Roscoe,H.K.,andPlane,
J.M.C.:Boundarylayerhalogensincoastalantarctica,Science,317,348,doi:10.1126/science.1141408,
2007b.Saiz-Lopez,A.,Plane,J.M.C.,Mahajan,A.S.,Anderson,P.S.,Bauguitte,S.J.-B.,Jones,A.E.,Roscoe,
laH.yeK.,rhalogensSalmon,ovR.erA.,coastalBloss,anW.tarctica:J.,Lee,J.implicationsD.,andforHeard,O3,D.HOxE.:,OnNOxtheandvtheerticalHgdistributionlifetime,Aoftmos.boundaryChem.
Phys.Discuss.,7,9385–9417,2007c.
Saiz-Lopez,A.,Plane,J.M.C.,Mahajan,A.S.,Anderson,P.S.,Bauguitte,S.J.-B.,Jones,A.E.,Roscoe,
H.K.,Salmon,R.A.,Bloss,W.J.,Lee,J.D.,andHeard,D.E.:Ontheverticaldistributionofboundary
layerhalogensovercoastalantarctica:implicationsforO3,HOx,NOxandtheHglifetime,Atmospheric
ChemistryandPhysics,8,887–900,2008.

184

Sander,R.:(asreferee)Interactivecommenton”Amechanismforbiologically-inducediodineemissionsfrom
sea-ice”byA.Saiz-LopezandC.S.Boxe,AtmosphericChemistryandPhysics,8,S602–S603,2008.
Sander,R.andCrutzen,P.J.:Modelstudyindicatinghalogenactivationandozonedestructioninpolluted
airmassestransportedtothesea,J.Geophys.Res.,101,9121–9138,1996.
Sander,R.,Kerkweg,A.,Jöckel,P.,andLelieveld,J.:Technicalnote:Thenewcomprehensiveatmospheric
chemistrymoduleMECCA,Atmos.Chem.Phys.,5,445–450,2005.
Sander,R.,Burrows,J.,andKaleschke,L.:Carbonateprecipitationinbrine-apotentialtriggerfortropo-
sphericozonedepletionevents,AtmosphericChemistryandPhysics,6,4653–4658,2006a.
Sander,R.,Gromov,S.,Harder,H.,Jöckel,P.,Kerkweg,A.,Kubistin,D.,Riede,H.,Taraborrelli,D.,and
Xie,Z.-Q.:TheatmosphericchemistryboxmodelCAABA/MECCA-3.0,GeoscientificModelDevelop-
2009.Discussions,tmenC.Sander,E.,S.P.Kurylo,,FinlaM.J.,yson-Pitts,Molina,B.M.J.,J.,Friedl,MoR.ortgat,R.,G.K.,Golden,Orkin,D.M.,V.L.,Huie,RaR.E.,vishankara,Keller-Rudek,A.R.,andH.,KWine,olb,
P.H.:ChemicalKineticsandPhotochemicalDataforUseinAtmosphericStudies,JPLPublication06-2,
EvaluationNumber15,JetPropulsionLaboratory,Pasadena,2006b.
Sandu,A.andSander,R.:Technicalnote:Simulatingchemicalsystemsinfortran90andmatlabwiththe
kineticpreprocessorKPP-2.1,AtmosphericChemistryandPhysics,6,187–195,2006.
Schall,C.,Laturnus,F.,andHeumann,K.G.:Biogenicvolatileorganoiodineandorganobrominecompounds
releasedfrompolarmacroalgae,Chemosphere,28,1315–1324,1994.
Schönhardt,A.,Richter,A.,Wittrock,F.,Kirk,H.,andBurrows,J.P.:Firstobservationsofatmospheric
iodineoxidecolumnsfromsatellite,Geophys.Res.Abstr.,9,2007.
Schönhardt,A.,Richter,A.,Wittrock,F.,Kirk,H.,H.,O.,Roscoe,H.K.,andBurrows,J.P.:Observations
ofiodinemonoxidecolumnsfromsatellite,Atmos.Chem.Phys.,8,637–653,2008.
Schröter,W.,Lautenschläger,K.-H.,andBibrack,H.:TaschenbuchderChemie,VerlagHarriDeutsch,13
1988.edn.,Shefov,N.N.:Spectroscopic,photoelectric,andradarinvestigationsofauroraeandthenightglow,Izd.
1959.1,Nauk,ad.AkySimpson,earsea-iceW.R.,contactCarlson,predictsD.,bromineHönninger,monoG.,xideDouglas,(BrO)T.levA.,elsatSturm,barroM.,w,Peroalaskvicah,betterD.,andthanpPlatt,otenU.:tialFirst-frost
flowercontact,AtmosphericChemistryandPhysics,7,621–627,2007a.
Simpson,W.R.,vonGlasow,R.,Riedel,K.,Anderson,P.,Ariya,P.,Bottenheim,J.,Burrows,J.,Carpenter,
L.J.,Frieß,U.,Goodsite,M.E.,Heard,D.,Hutterli,M.,Jacobi,H.-W.,Kaleschke,L.,Neff,B.,Plane,
J.,Platt,U.,Richter,A.,Roscoe,H.,Sander,R.,Shepson,P.,Sodeau,J.,Steffen,A.,Wagner,T.,and
Wolff,E.:Halogensandtheirroleinpolarboundary-layerozonedepletion,AtmosphericChemistryand
2007b.4375–4418,7,ysics,PhSinnhBurrouber,ws,J.B.-M.,P.:GlobalRozanov,observA.,ationsSheoofde,N.,stratosphericAfe,O.T.,bromineRichter,monoA.,xideSinnhfromuber,SCIAMAM.,CHY,Wittrock,GeophF.,ysicaland
ResearchLetters,32,doi:10.1029/2005GL023839,2005.
Solomon,S.,Schmeltekopf,A.L.,andSanders,R.W.:Ontheinterpretationofzenithskyabsorption
measurements,JournalofGeophysicalResearch,92,8311–8319,1987.
Solomon,S.,Garcia,R.R.,andRavishankara,A.R.:Ontheroleofiodineinozonedepletion,J.Geophys.
Res.,99,20491–20499,1994.
Spietz,P.,GómezMartín,J.C.,andBurrows,J.P.:Spectroscopicstudiesofthei2/o3photochemistry,
Part2:ImprovedspectraofiodineoxidesandanalysisoftheIOabsorptionspectrum,J.Photochem.
Photobiol.A,176,50–67,doi:10.1016/j.jphotochem.2005.08.023,2005.

185

Spreen,G.,Kaleschke,L.,andHeygster,G.:Seaiceremotesensingusingamsr-e89ghzchannels,J.Geophys.
2008.doi:10.1029/2005JC003384,113,Res.,Stöcker,H.:TaschenbuchderPhysik,VerlagHarriDeutsch,2ndedn.,1995.
Stolarski,R.S.andCicerone,R.J.:Stratosphericchlorine:apossiblesinkforozone,Can.J.Chem.,52,
1974.1610–1615,Stutz,J.,Pikelnaya,O.,Hurlock,S.C.,Trick,S.,Pechtl,S.,andvonGlasow,R.:DaytimeOIOintheGulf
ofMaine,Geophys.Res.Lett.,34,L22816,doi:10.1029/2007GL031332,2007.
Thomas,D.N.andDiekmann,G.S.,eds.:SeaIce:Anintroductiontoitsphysics,chemistry,biologyand
geology,BlackwellPublishingCompany,2003.
Tokarczyk,R.andMoore,R.M.:Productionofvolatileorganohalogensbyphytoplanktoncultures,Geo-
physicalResearchLetters,21,285–288,1994.
Truesdale,V.W.,Canosa-Mas,C.E.,andW.,L.G.:Disproportionationandreductionofmoleculariodine
addedtoseawater,MarineChemistry,51,55–60,1995.
Tucceri,M.E.,Hölscher,D.,Rodriguez,A.,Dillon,T.J.,andCrowley,J.N.:Absorptioncrosssectionand
photolysisofOIO,PhysicalChemistryChemicalPhysics,8,834–846,2006.
Vasilkov,A.P.,Joiner,J.,Gleason,J.,andBhartia,P.:OceanRamanscatteringinsatellitebackscatterUV
measurements,Geophys.Res.Let.,10,10,2002.
Vogt,R.,Sander,R.,vonGlasow,R.,andCrutzen,P.J.:Iodinechemistryanditsroleinhalogenactivation
andozonelossinthemarineboundarylayer:Amodelstudy.,J.Atmos.Chem.,32,375–395,1999.
Volkamer,R.,Spietz,P.,Burrows,J.,andPlatt,U.:High-resolutionabsorptioncross-sectionofglyoxalin
theuv/visandirspectralranges,J.Photochem.Photobiol.,172,2005.
vobnGlasooundaryw,layR.,er.1.Sander,Cloud-freeR.,Bott,MBL,A.,J.andGeophys.Crutzen,Res.,P.107,J.:4341,Modelinghalogendoi:10.1029/2001JD000942,chemistryinthe2002.marine
Vounradiativtas,eM.,transferRozanoinv,V.earth’sV.,andatmosphere,Burrows,J.J.QuanP.:t.SpRingectrosc.effect:Radiat.ImpactTofransfer,rotational60,raman943–961,1998.scatteringon
Vountas,M.,Richter,A.,Wittrock,F.,andBurrows,J.P.:Inelasticscatteringinoceanwateranditsimpact
ontracegasretrievalsfromsatellitedata,AtmosphericChemistryandPhysics,3,1365–1375,2003.
Vrekoussis,M.,Wittrock,F.,Richter,A.,andBurrows,J.P.:Temporalandspatialvariabilityofglyoxalas
observedfromspace,AtmosphericChemistryandPhysics,9,4485–4504,2009.
Wbada,oundaryR.,laBeames,yerJ.usingM.,acandavityOrr-Ewing,ring-downA.spJ.:ectrometer,MeasurementJournalofIOofAradicaltmosphericconcentrationsChemistryin,the58,marine69–87,
2007.doi:10.1007/s10874-007-9080-z,Wagner,T.andPlatt,U.:Satellitemappingofenhancedbroconcentrationinthetroposphere,Nature,395,
1998.486–490,Wayne,R.P.:ChemistryofAtmospheres,OxfordUniversityPress,3rdedition,2000.
Wennberg,P.O.,Brault,J.W.,Hanisco,T.F.,Salawitch,R.J.,andMount,G.H.:Theatmospheric
columnabundanceofIO:implicationsforstratosphericozone,J.Geophys.Res.,102,8887–8898,1997.
Whalley,L.K.,Furneaux,K.L.,Gravestock,T.,Atkinson,H.M.,Bale,C.,Ingham,T.,Bloss,W.J.,
andHeard,D.:Detectionofiodinemonoxideradicalsinthemarineboundarylayerusinglaserinduced
fluorescencespectroscopy,J.Atmos.Chem.,58,19–39,doi:10.1007/s10874-007-9075-9,2007.
WHO:IodineDeficiencyinEurope:Acontinuingpublichealthproblem.Andersson,M.,anddeBenoist,B.,
andDarnton-Hill,I.,andDelange,F.(eds.),WorldHealthOrganization,ISBN978-92-4-159396-0,2007.

186

Wittrock,F.:Theretrievalofoxygenatedvolatileorganiccompoundsbyremotesensingtechniques,Ph.D.
thesis,UniversityBremen,2006.

Wittro(IO)ckab,ovF.,eSpitsbMüller,R.,ergen,RichGeophter,ys.A.,BoRes.vLett.,ensmann,27,H.,and1471–1474,Burrows,2000.J.P.:Measurementsofiodinemonoxide

Wittrock,F.,Oetjen,H.,Richter,A.,Fietkau,S.,Medeke,T.,Rozanov,A.,andBurrows,J.P.:MAX-DOAS
measurementsofatmospherictracegasesinNy-Ålesund-Radiativetransferstudiesandtheirapplication,
Atmos.Chem.Phys.,4,2004.

WMO,Pro2006:ject-ReportScienNo.tific50,WAssessmenorldtofMeteorologicalOzoneDepletion:Organization,2006,finalGlobalreleaseOzonepublishedResearcFhebuaryand2007.Monitoring

laZingler,yerJ.IO,J.andGeophPlatt,ys.U.:IRes.,odine110,oxideinthedoi:10.1029/2004JD004993,deadseavalley:2005.Evidence

for

inorganic

sources

of

oundaryb

187

rungErklä

Hiermiterkläreich,dassichdieArbeitohneunerlaubtefremdeHilfeangefertigthabe,dassich

weiterhinkeineanderenalsdieangegebenenQuellenundHilfsmittelbenutztunddiedenbenutzten

Werkenwörtlichoderinhaltlichentnommenen

reb03.SeptemBremen,

njaA

dtSchönhar

0092

Stellen

als

hesolc

htlicennk

thgemac

e.hab

wledgementsknocA

Lookingbackoverthepastyearsandthinkingofthepeopledirectlyorindirectlyconnectedtothis
work,thereareseveralpeopleIwouldliketothankspecificallyfortheirsupport,astheyhavemade
ossible.porkwthis

Prof.J.P.Burrowshasofferedmetheopportunityofworkinginhisresearchgroupasadoctoral
student.Thankyouverymuchforallyourgoodadvice,yourencouragementandtheinsightinto
discussions.tificscienyman

Prof.LarsKaleschkedankeichfürdiehilfsbereiteundfreundlicheÜbernahmedesZweitgutacht-
ens,insbesondereinAnbetrachtderterminlichenUmstände.ZusätzlichhatesindenletztenJahren
immerwiedersomancheinteressanteDiskussiongegeben.VielenDankauchdafür.

AndreasRichtermöchteichfürseinewesentlicheundunersetzlicheUnterstützungdanken.Esgibt
kaumjemanden,vondemichwissenschaftlichsovielgelernthabe.Außerdembewohnenwirzusam-
menmitJoanadasbesteBürodesInstitutsinsehrangenehmerundunterhaltsamerAtmosphäre.

IwouldliketothanktheentireDOAS-Group(presentandpast)whereIhavealwaysfeltcom-
fortable.Thankyouespeciallyforthepositiveatmosphere,themutualsupport,manyinteresting
andfruitfuldiscussions,andsomeDOAS-activitiesaswellasforyourfriendship.ThankyouAn-
dreas,Folkard,Henning,Joana,Annette,Mihalis,Enno,Frank,Mathias,AndreasHeckel,Andreas
Hilboll,Katsuyuki,andHilke.

VielenDankandieserStelleauchanAndreas,FolkardundJoanafürdasKorrekturlesendieser
eit.Arb

VielenKollegenausdemIUPausdenverschiedenenArbeitsgruppen,ausderVerwaltungundder
Systemadministrationbinichsehrdankbarfürihrewissenschaftliche,organisatorischeundcomput-
erbezogene,sowiemoralischeUnterstützung.

AucvielmalshwennderdieWLaberkstattorarbdeseitenFB1infürdiediescArbhriftliceitenheArbaneitderkeinenOptikfürEinzugdasgeplangefundentehaben,Flugzeuggerät.dankeich

RolfSanderdankeichfürdieBereitstellungdesCAABA/MECCAModellsunddiegeduldigeBeant-
wortungmeinervielenFragen.

HowardRoscoedeservesmygratitudeforsomefruitfuldiscussions.

Iamgratefultothefollowingpeopleandorganisationsfortheprovisionofdataproducts:
-LarsKaleschkeandGunnarSpreenfortheiceconcentrationmaps
-AstridBracherforthediatomconcentrationmap
-MarcoVountasandWolfgangLotzforPMDcloudandsurfacetypedatapriortopublication
-AndreasRichterandMathiasBegoinforBrOdata
-AndreasRichterfortheGOME-2resultsofthewatereffectfitting
-HilkeOetjenandFolkardWittrockfortheIOdatainNy-Ålesund
-AlfonsoSaiz-LopezandJohnPlanefortheground-baseddatafromtheCHABLIScampaign
-ESAandtheDLRforthelevel1datafromSCIAMACHY
-SHOMfortidalheightdatainthemid-latitudes
-NASAfortheSeaWIFSchlorophyllmaps

DerChristian.größteEureDankLiebgehetaundnUnmeineFterstützungamilie-sindmeinedasWicEltern,htigstemeinenfürmicBruderh.mitseinerFamilieund

VielenbauendeWDank,ort.IhrMamaseidundPwirklicapa,hfürimmerEurefürbmicheständigeda.undliebevolleUnterstützungundjedesauf-

GroßerBruder,vielenDankdassichmichjederzeitaufDichundDeineUnterstützungverlassen
ann.k

Christian,dassDudabistundimmerzumirhälst,gibtvielemeinenSinnundmirimmerneue
Kraft.EinsiebenfachesDankeschön.