teb301533/teb078 Tellus.cls October1,2003 14:29 Tellus(2003),55B,993–1006 Copyright(cid:2)C BlackwellMunksgaard,2003 PrintedinUK.Allrightsreserved TELLUS Analysis of the Algerian severe weather event in November 2001 and its impact on ozone and nitrogen dioxide distributions ∗ ByW.THOMAS1 ,F.BAIER2,T.ERBERTSEDER2 andM.KA¨STNER2,1Institutfu¨rMethodikderFern- erkundung (IMF), and 2Deutsches Fernerkundungsdatenzentrum (DFD), Deutsches Zentrum fu¨r Luft- und Raumfahrt(DLR),P.O.Box1116,D-82234Wessling,Germany (Manuscriptreceived1July2002;infinalform14April2003) ABSTRACT AnanalysisofthesevereweathereventinNovember2001overthewesternMediterraneanispresented focusingonsatellite-basedtracegasmeasurementsfromtheGlobalOzoneMonitoringExperiment (GOME)onboardtheEuropeanRemoteSensingSatellite(ERS-2).Thisstudyissupplementedby asynopticanalysisandsimulationsofthethree-dimensionalstratosphericchemicaltransportmodel ROSE.ArcticairmassesmovedrapidlyfromScandinaviatotheIberianpeninsulaandweremixedwith subtropicalairoverthestillwarmMediterraneanSea.Thiscausedseverethunderstormsandextreme rainfallalongthecoastsofMoroccoandAlgeriaandlaterontheBalearicIslands.Associatedwiththe meridionaltransportanintrusionofstratosphericairbelow3kmabovesealevelwasobserved.The maximumpotentialvorticity(PV)derivedfromUKMeteorologicalOfficeanalysisdatawasabout9.3 potentialvorticityunits(pvu)at330Kattheequatorwardpositionof35◦N.Theseveryhighvalueswent alongwithremarkablyenhancedtotalozonelevelsobtainedfromGOMEbackscattermeasurements ofcollocatedGOME/ERS-2overpasses.FurtherinvestigationofGOMEdatashowedunusuallyhigh levelsofnitrogendioxide(NO2)abovethewesternMediterranean.Wepresentanewmethodtoderive thetroposphericcontentofnitrogendioxide(NO2)fromacombinationofsatellitemeasurementsand resultsofachemicaltransportmodel.Weshowthatabouttwo-thirdofthetotalatmosphericcontentof nitrogendioxideintheobservedplumeisfoundinthetroposphere,duetolightningactivity,advection andverticaltransportinthethunderstormsfromtheplanetaryboundarylayer(PBL)toatmospheric levelsaboveclouds. 1. Introduction andnitrogendioxideduringthisweatherevent,both derivedfromGOMEnadirbackscattermeasurements, From9to11November2001,aseriesofheavythun- drewourattentiontoitandweanalyseditstemporal derstormswasobservedoverMorocco,Algeriaandthe andspatialevolutioninthestratosphereandthetro- BalearicIslandsthatcausedseveredamagetothein- posphere.Wefurtherusedthethree-dimensional(3D) frastructure,mainlyduetoflooding.Althoughheavy chemical-dynamicaltransportmodelROSE(Roseand rain and cyclones are typical events in the western Brasseur,1989)tosimulatetransportandthenitrogen Mediterraneaninwinter,morethan750peopledied dioxide background distribution in the stratosphere intheflood(“elhemla”),whichwasthesecondworst andtocomparethelatterwithGOMEmeasurements. naturaldisasterinAlgeriaduringthelast40years.The Anewmethodwasdevelopedthatcombinessatellite unusualpatternofthespatialdistributionoftotalozone observationsandROSEresultsinordertoderivethe troposphericnitrogendioxidecontent. ∗Correspondingauthor. This severe weather event originated from an e-mail:[email protected] arctic cold front along which air masses moved Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 994 W.THOMASETAL. within two days over several thousand kilometres 2. Data southward from Scandinavia over the Iberian penin- sulatoNorthAfrica,wherecyclogenesisstarted.Cy- 2.1. Tracegasmeasurements clogenesis is a well known process for cut-off lows Theimpactofasevereweathereventonthechem- whenairmassesofrelativelyhighpotentialvorticity ical composition of the atmosphere is analysed us- (PV) rapidly push down southwards from polar re- ingbackscattermeasurementsfromtheGlobalOzone gions.ApproachingtheaxisofthetroughinaRossby MonitoringExperiment(GOME)onboardtheERS- waveaconversionofstaticstabilityintoabsolutevor- 2 satellite (see e.g. Burrows et al., 1999 and refer- ticitytakesplacebystretchingoftheverticalaircol- ences therein). An enhanced version of the opera- umn.Theformationofacut-offlowstartsandcoin- tionalGOMEDataProcessor(GDP)asdescribedin cidesinthiscasewiththedryintrusionofstratospheric Loyolaetal.(1997)wasusedtoderivetotalcolumnre- air,whichoccursfrequently(Kentarchosetal.,1998). sultsofozoneandnitrogendioxide.TheGOMEspec- TheseprocessesaretriggeredbyPVanomalies,asre- trometerisabletomeasuretheatmosphericcolumnar centlydescribedbyBluestein(1993).Duetothelarge- contentofanumberofminortraceconstituents(e.g. scaleRossbywavepatternozone-richairinthelower ozone, nitrogen dioxide, bromine monoxide, sulfur stratosphereistransportedsouthward. dioxide,formaldehyde,chlorinedioxide)between240 Nitrogen oxides play a key role in the formation and793nm,withspectralresolutionsvaryingfrom0.2 of tropospheric ozone, a photochemical process that to0.4nm.GOMEisanadir-lookingacross-trackscan- remarkably influences the chemical composition of ninginstrumentwithatypicalfootprintsizeofabout theatmosphere.Thequantitativecontributionsofboth 320×40km2. naturalandanthropogenictroposphericsourcesofni- Since GOME pixels cover a relatively large area, trogen dioxide to the total NO budget are not very 2 the footprints will often be partially or totally cov- well known, due to sparse ground-based measure- ered by clouds. Clouds are opaque in the UV/VIS ments.Here,satellite-basedmeasurementsmaycon- spectral range (except optically thin cirrus clouds), tributesignificantlytoprovidetroposphericnitrogen whichneedtobetakenintoaccountiftracegastotal dioxide levels on a global scale (Lauer et al., 2002; columnsareretrieved.ThecloudcoverageofaGOME Velders et al., 2001; Martin et al., 2002). Recent footprintisdeterminedbyalinearleast-squaresfitof airbornemeasurementssubstantiatedthesupposition GOMEspectratosimulatedspectrainandaroundthe that considerable contributions originate from light- oxygenA-bandbetween758and778nm.Weusea ning in thunderstorms and vertical transport of pol- relationship between the measured absorption depth lutedairmasses(i.e.NO -richair)fromtheplanetary 2 in the oxygen A-band, the cloud optical thickness, boundary layer (PBL) to higher atmospheric levels the cloud-top height and the fractional cloud cover (Huntrieseretal.,2002;Brunneretal.,2001).Atthe ofaGOMEpixelasdescribedindetailbyKuzeand tropopause level further contributions from air traf- Chance(1994).CloudcoverageresultsfromGOME fic have also to be taken into account (Schumann, can be lower than corresponding results from other 1994).IntheupperstratosphereNO ismainlypro- 2 satellitesensors(KoelemeijerandStammes,1999)but duced via atomic oxygen from nitrous oxide (N O) 2 comparereasonablywellifopticallythickcloudsare and is subsequently transported poleward into the present,asitismainlythecaseinourstudy(seee.g. lower stratosphere. There, denitrification and trans- Fig.2). portintothetropospherecontributetothemainloss ThecoreelementoftheretrievalpackageisaDOAS processes.ThedailyaveragedstratosphericNO col- 2 (DifferentialOpticalAbsorptionSpectroscopy)fitting umn is dominated by available solar insolation and technique that involves a multi-linear regression of showsapronouncedlatitudinaldependency,whiletro- GOME-measured optical densities against a number posphericNO levelsarehighlyvariableintimeand 2 of reference spectra. It provides trace gas column space. amounts along the viewing path of the instrument The trace gas retrieval technique and the ROSE (Burrowsetal.,1999).Thedifferentialabsorptionof model are described in section 2; in section 3 we trace species along the absorption path is modelled present the synoptic analysis of the troposphere and accordingtoBeer’slaw stratosphere,the3DresultsofROSE,andthederiva- tionofthetroposphericnitrogendioxidecontent.Our conclusionisgiveninsection4. dI(λ)=−I(λ)σ(λ)C(s)ds (1) Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 ALGERIANSEVEREWEATHEREVENT 995 where the incremental change of intensity dI(λ) at Acommonwayofcalculatingtheverticalcolumn wavelengthλthroughaslantpathdistancedsispro- density(VCD)ofaspecies(herenitrogendioxide)is portional to the absorption cross-section σ(λ) times tocalculatefirstarepresentativeslantcolumnSC total the intensity I(λ) and the absorber column amount byaddingthefittedslantcolumnSCandaclimato- C(s).Whenthereareseveralweakabsorbers,thecon- logicalslantcolumncontentbelowclouds(ifpresent), tributionsareadditive,providedthatthetotaloptical followedbyadivisionthroughthetotalairmassfac- density remains small and therefore saturation does torAMF whichistheweightedaverageofclearsky total notoccur.Theozoneretrievalwasperformedbetween andcloudyAMFs.Aso-calledghostverticalcolumn 325and335nm,whiletheNO retrievalusedafitting (GVC) below clouds is derived from climatological 2 windowbetween425and450nm.Detailsaboutthe trace gas profiles. The vertical column density now spectralfittingcanbefoundinBurrowsetal.(1999) reads: andThomasetal.(1998). SC SC+ f ×GVC×AMF The resulting trace gas slant columns are then VCD= AMFtotal = f ×AMFc+(1− f )×AMc F converted to geometry-independent vertical column total c c c g (3) amountsthroughdivisionbyappropriateairmassfac- tors (AMF). Air mass factors describe the enhance- wheref isthecloudcoveragebetween0and1,SCthe c mentoftheabsorptionofagiventracegasduetoslant fitted slant column, GVC the ghost vertical column, pathsofincidentlightintheatmosphere: and AMF and AMF the air mass factors down to g c τ groundandtocloud-top. AMF= slant (2) Recent studies (Lambert et al., 1999; 2000) com- τ vert paringGOMEandTOMSdataagainstground-based whereτ ,τ aretheslantandtheverticaloptical measurementsindicatethattheagreementofGOME slant vert densities,respectively.Theslantopticaldensities,i.e. totalozonecolumnsiswithin2–4%forsolarzenith theAMFs,arederivedfromradiativetransfercalcula- anglesbelow70◦,whichisthecaseinourstudy.Total tionsanddonotrequireinputfromGOMEmeasure- nitrogendioxidecolumnsderivedfromGOME,other ments other than viewing geometry and geolocation satellite instruments (HALOE, POAM) and ground- information. However, AMFs depend further on the basedmeasurementsagreeingeneralwithin5–20%, albedo of the underlying surface and the altitude of but within ±5×1014 molecules cm−2 in areas of thereflectingsurface,whichcanbeeithertheground lowNO intheplanetaryboundarylayerandwithin 2 orthecloud-top,andatmosphericprofilesoftemper- ±8×1014 moleculescm−2 inareasofverylowslant ature,pressureandtracegasconcentration.Radiation columnsofnitrogendioxide.Recentstudies(Richter propagationthroughcloudsisnotconsidered.Instead, andBurrows,2002;Martinetal.,2002)showedthat clouds are treated as reflecting boundaries in the at- GOMEnitrogendioxidemeasurementspartiallysuf- mosphere. ferfrominstrumentalartefactsthatmaycauseatem- Ozone AMFs were calculated on the basis of the porallyvaryingoffsetintheslantcolumnsupto2× TOMS V7 ozone profile climatology (Klenk et al., 1015 molecules cm−2 depending on the time of the 1983;Chuetal.,1989)usingLIDORT(Spurretal., year.Thesestructures,however,appeartobesimilar 2001),andareobtainedfromlook-uptablesbyneural fromyeartoyear.ForNovember2001,aslightunder- networktechniques.Thetotalozonecontentisderived estimationofretrievedNO slantcolumnsaround4× 2 fromaniterativeprocedurethatsearchesforthebest 1014moleculescm−2canbeestimatedusingtheoffset suitedozoneprofile,i.e.theAMFasfunctionofthe determinedfor1998/1999byMartinetal.(2002). integratedprofilecontent(andanumberofothergeo- physical parameters, as mentioned above). A differ- 2.2. ThechemicaltransportmodelROSE entapproachisfollowedfornitrogendioxideAMFs, which are calculated in two steps. A fast radiative The spatial background distribution and the transfermodelcalculatesAMFson-line,considering temporal evolution of stratospheric ozone and ni- onlysinglescatteringeffects.Themultiplescattering trogen dioxide are simulated using the 3D global isthenaccountedforbymultiplicativecorrectionfac- chemical-transportmodelROSEasdescribedindetail torsthatwerederivedfromradiativetransfersimula- in Rose and Brasseur (1989), Granier and Brasseur tions(Rozanovetal.,1997),restoredfromalook-up (1991)andRieseetal.(1999).Themodelcoversthe table. relevant gas-phase stratospheric chemical processes. Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 996 W.THOMASETAL. Heterogeneous processes on polar-stratospheric Thisdatasetdescribesclimatologicalzonalmean clouds and on sulfuric aerosols are also included distributionsofasub-setofspeciesrelevantforstrato- in the model. It accounts for about 100 reactions, sphericconditions.Themodel’stropopauseisdefined including oxygen, hydrogen, carbon, nitrogen, by a potential vorticity of 1.6 pvu (see section 3.2) chlorine and bromine species. The chemical rate or a potential temperature of 380 K for the tropics. constants and cross-sections are taken from Sander In order to investigate the impact of the tropopause et al. (2000). Photolysis rates are derived from a heightmodelrunswerealsoperformedforadifferent look-uptabledependingonsolarzenithangle,ozone tropopauseheightcorrespondingto3pvu(seesection columnandaltitude.Thechemicalrateequationsare 3.4).Themodelrunscoveringtheperiod6–14Novem- solved by considering a chemical equilibrium state ber2001wereinitialisedwithdatafromtheDFDnear- fortheshort-livedspecies(e.g.ClO,NO,HO,BrO). real-time(NRT)service(http://auc.dfd.dlr.de/ROSE). Asemi-implicitschemeisusedfortheintegrationof ThisNRTservicedeliversverticalozoneprofilesusing the more stable reactants (e.g. HNO , N O, NH ). GOMEtotalozonecolumndensitiesthatareassimi- 3 2 4 All short-lived species are grouped and integrated latedintoROSEonanoperationalbasis. using families, e.g. ClO =Cl+ClO, NO =NO+ x x NO . All long-lived species are transported using 2 3. Results the semi-Lagrangian scheme of Smolarkiewicz and Rasch(1991).Windandtemperaturefieldsarederived 3.1. Synopticanalysis from24-hanalysesoftheUKMeteorologicalOffice (UKMO) following Swinbank and O’Neill (1994), The meteorological situation during the weather whichareavailableforpressurelevelsfromgroundup event is characterised by strong surface winds and to0.3hPa.Thisdatasetdefinesaconsistentsynoptic heavyrainfall.Figure1(leftpanel)showsanintense state using satellite-based temperature soundings upperleveltroughthatpushesfartotheSouthwestof and radiosonde observations assimilated in a global EuropeandmeetstheAlgeriancoastfromweston9 circulationmodel(GCM). November2001.TheUKMOanalysisofthetempera- ForthisstudyweuseaROSEversionwitha5.6◦× tureandthegeopotentialheightisgivenatthe464hPa 5◦longitude–latitudesphericalgridon43log-pressure level(∼6km).TheMETEOSAT-7IR-imagefromthe levelsbetween0and60kmaltitude,resultinginaver- sameday(Fig.2,leftpanel)showsthatthecloudsys- ticalstepsizeof1.3km.Inthemodel’stroposphere tems of the frontal zone extend from Finland across all chemical species are relaxed to 2D model data CentralEuropetoSpain.Thecoldfrontmadethesur- (Brasseuretal.,1995)withatimescaleof10d. facetemperaturesinAndalusia(Spain)dropatmidday Fig.1. UKMOanalysisofthetemperatureinK(greyscale)andthegeopotentialheightingpm(isolines)at464hPa(∼6km) at12:00UTCfor9November2001(leftpanel)and10November2001(rightpanel). Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 ALGERIANSEVEREWEATHEREVENT 997 Fig.2. METEOSAT-7IRdataat12:00UTCfor9November2001(leftpanel)and10November2001(rightpanel);A, Algiers. from25to18◦Candtofrostatnight.Heavyrainfall overpass time over the area is around 10:20 h local beginsattheAlgeriancoastonlate9November2001 time(herecloseto10:20UTC)oneachdayduetothe and ends at about noon on 10 November 2001. Cy- sun-synchronous polar orbit of ERS-2. High gradi- cloniccurvatureandshearvorticitydefinethepositive entsintotalozonecolumnfrompixeltopixel(several vorticityadvection(PVA)attheAlgeriancoast.This 10DU)indicatestrongdynamicactivityoverEurope. PVAimpliesadeepeningofthelow-pressuresystem Asearlyasin1950Reedshowedthatthetotalozone and a cut-off low developed with maximum poten- columnisanexcellenttracerforupperlevelweather tialvorticityinthatarea(Fig.1,rightpanel).Onits systems modulated by tropopause and lower strato- east side very warm lower air from the tropics was sphere variability. A well suited synoptic diagnosis mixed with higher cold air of arctic origin, leading quantityisthecorrespondingisentropicPVat330K to a highly unstable air mass and initiating a cyclo- (Hoskinsetal.,1985)whichisdepictedforeachday genesis with a tropical warm core which is unusual on the right panels of Fig. 3. PV is a conservative attheselatitudes.Twoprocessesintensifiedthecon- quantityinfrictionlessandadiabaticflowandtheval- vectivedevelopment:(1)thecoldmaritimearcticair uesonthe330Kisentropecorrelatewellwiththeto- crosses over the still warm Mediterranean, where it talozonecolumnatmidlatitudes(Hoodetal.,1999). picksupmoistureandmeetsmaritimesubtropicalair Theessentialphysicsoftherelationshipbetweentotal and(2)thestrongsurfacewindsfromnorth-westto- ozoneandvorticityisdescribedinVaughanandPrice wardsthemountainousAfricancoastcausedintense (1991). orographicrainfall.Figure2showsthecloudsrotating Atthebeginningoftheinvestigatedperiod,weob- aroundAlgiersduetothecut-offlowprocess.Onboth servelowozonelevelsbelow200DUoverNorthern days,9and10November2001,extraordinaryrainfall Europeon9and10November2001forminganozone ratesof126Lm−2in12hand136Lm−2in6hwere mini-hole(Figs.3aandc).Itisaccompaniedbyahigh- reported which led to the flooding disaster in Alge- pressuresystemwhichischaracterisedbyanelevated ria.Theregionalmaximumrainratein6hinAlgeria tropopause and a negative PV anomaly and governs wasmorethanthe30yearsmonthlymeanNovember the large-scale motion field (Fig. 3b). Hence, on its rainfallrateof129Lm−2. frontside(east)anupperleveltroughisformingover Scandinavia and the Baltic. It moves eastwards and rapidlyexpandsinasouth-westwarddirectionleading 3.2. OzoneandPVanalysis toenhancedozonelevelsoverSpain(Fig.3c).While ThetotalozonedistributionasderivedfromGOME the upper-level trough deepens, the tongue of high backscattermeasurementsbetween9and11Novem- PV elongates and thins with a pronounced NE–SW ber2001isshowninFig.3,leftpanels.TheGOME axis (Fig. 3d). The formation is due to equatorward Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 998 W.THOMASETAL. Fig.3. GOME-measuredozonedistributionon9November2001(a),10November2001(c),11November2001(e)and corresponding isentropic potential vorticity on the 330 K surface derived from UK Meteorological Office stratospheric analysesforthesamedaysat12:00UTC(b,d,f). Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 ALGERIANSEVEREWEATHEREVENT 999 Rossbywavebreaking(McIntyreandPalmer,1985). mosphere. The dynamic tropopause can be defined From10to11November2001,thestructureformed asanisosurfaceofpotentialvorticity.Hoerlingetal. byhighPVlevelsbreaksupandthetipofthetongue (1991)discusseddifferentthresholdvaluesfrom1to separates, rolling up cyclonically, forming a cut-off 3.5pvuforthedefinitionofthedynamictropopause lowcentredoverMorocco(Fig.3f).Thecoreshows whiletheWMOrecommendedthe1.6pvuisolineat high PV values of about 9.3 pvu (1 pvu = 10−6 K midlatitudes (WMO, 1985). According to Price and m2 kg−1 s−2), indicating a reservoir of polar strato- Vaughan(1993),thisisreasonablewhenstudyingcut- sphericairmassesataverypronouncedsouthernloca- offlows,asitisthecasehere.However,outsidethe tionof35◦N.AssociatedwiththeequatorwardRossby cut-offlowthe3pvuisolinemaybeamorerepresen- wavebreakingeventweobservehightotalozonecol- tativevalue.Inourcasestudythe3pvuisolineisabout umn values over Southern Spain, Morocco, Algeria 100hPahigherthanthe1.6pvuisolineintheareawith andtheMediterraneanSea(Figs.3cande).Theto- thecut-offlow,whiletheeffectislesspronouncedout- tal ozone content at these low latitudes increases by sidewherechangesarebelow25hPa.Theimpactof about 100 DU, from 280 DU on 8 November 2001 differenttropopauseheightsisdiscussedinmorede- to380DUon10November2001.On11November tailsinsections3.4and3.5.Thetropopauselowering 2001themaximumtotalozonehasdecreasedagainto withintheupperleveltroughleadstoanincreaseof 355DU. thestratosphericairreservoirandcontributestoanin- Inordertoinvestigatetheequatorwardlocationof creaseofthetotalozonecolumnamounts.Bymeans thePVmaximumfromaclimatologicalandstatisti- ofthePVasastratospherictracer,thefoldingshows cal point of view, 11 years of UKMO analysis data isentropicflowtoitswesternflank,indicatingstrato- wereexamined.Potentialvorticitywasdeterminedon sphericintrusion.AsdepictedinFig.3dthetipofthe adailybasisfrom1992to2002at35◦Nandforalllon- equatorwardpointingPVstructurebrokeawayon10 gitudesonthe330Kisentrope.DuringthisperiodPV November2001,formingacut-offlow.Forthatday, valuesreachedasimilarlevelonlyonfourdays,which Fig.4(leftpanel)showsanorth–southcross-section underlinesthestrongmeridionaltransportduringthis ofPVandpotentialtemperaturealongtheGreenwich outstandingweatherevent. meridianfrom20◦ to60◦N,basedonanalysesofthe Besideshorizontaltransportphenomenawealsoex- EuropeanCentreofMediumRangeWeatherForecast amined changes in the vertical structure of the at- (ECMWF). Fig.4. PV north–south cross-sections for 10 November 2001 at 0◦W (left panel) along the Greenwich meridian and METEOSAT-7 water vapour image (right panel) of 10 November 2001 11:30 UTC. Potential vorticity isocontours are inwhitewhileisentropesareinblack.The1.6pvuisoline(thickblackline,leftpanel)denotesthedynamicaltropopause.The nextisoline(white)correspondsto3pvuandthedifferenceto1.6pvuisshadedinbrightgrey.Thegeographicalposition ofAlgiersismarkedbytheAat36◦Nintheleftpanel.TheGreenwichmeridianat0◦Wismarkedbytheblacklineinthe METEOSAT-7image(Source:UKMetOffice). Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 1000 W.THOMASETAL. The cyclogenesis around Algiers (36◦N) is asso- 3.4. ROSEsimulationsofnitrogendioxide ciatedwithasignificantenfoldingofthetropopause ROSE simulations starting on 6 November 2001 wherestratosphericozone-richairintrudesto700hPa were performed to compare simulated stratospheric on10November2001andsplitsupintotwotongues nitrogendioxidedistributionswithcoincidentGOME inthecross-section(Fig.4,leftpanel).Thisisdueto observationsduringtheperiodofinterest.Themodels thecyclonicspinofthefoldaroundtheformingcut-off tropopauseisdefinedasdescribedinsection2.2using low,whichisconfirmedbytheMETEOSAT-7water the1.6pvuorthe380Kpotentialtemperatureisoline. vapourimagefrom10November2001at11:30UTC ThedifferencesbetweenGOMENO totalcolumnre- (Fig.4,rightpanel).Thedryintrusionofstratospheric 2 sults and modelled stratospheric results can then be airintothecycloneisapparent:itformsastreamerin- usedtoestimatethetroposphericNO contentwhich dicatedbyadarknarrowbandreachingstraightfrom 2 isfurtherdiscussedinsection3.5. RussiatoMoroccoandthenspinningcyclonically. Toseparatetheeffectoftransportfromlocalchem- ical transformations we analysed the temporal de- velopment of total ozone and a tracer (chemically 3.3. Nitrogendioxideobservations passive ozone, initialised with ozone values from 5 TheNO distributionderivedfromGOMEmeasure- November 2001) for a fixed model location at 0◦E, 2 ments does not show a similar temporal and spatial 40◦N.Airmasseswithhighozoneloadingweretrans- pattern as ozone and PV in the same period. How- ported southward, as confirmed by PV analysis (see ever,on10November2001,enhancedNO levelsare section 3.1). We observe a maximum increase of 2 observedalongtheCantabriccoast(Spain)andover the simulated total ozone column during 8 to 10 Morocco(alsooverMacedonia),whichcoincideswith November 2001 and these results compare qualita- night-timelightningactivity(Fig.5,upperpanels).En- tively well with coincident GOME observations. By hancedNO concentrationsintheuppertroposphere comparingthetemporalbehaviourofozonewiththat 2 causedbylightningactivityhaverecentlybeenmea- ofthetracerweestimatethatchemicaltransformations suredbyHuntrieseretal.(2002);theseplumesmay locallyonlycontribute5%tothesimulatedchangeof persistforseveralhours.Anotherareawithhigherni- thestratosphericozonecolumn. trogendioxidelevelsisvisibleoverFrancewhichcan Recent comparisons of ROSE results with obser- beattributedtotroposphericpollution.Here,GOME vations showed that the NO temporal behaviour is 2 cloud coverage results indicate a low cloudiness of reasonably well described by the model (Sen et al., below20%providingsatellitemeasurementsdownto 1998;Toon,1991).MaximumdailyNO columnsare 2 surfacelevels.On11November2001,duringandafter simulatedaftersunset(atsurfacelevels)beforeNO 2 heavythunderstorms,anNO2plumeisobservedover isconvertedmainlyintoitsnight-timereservoirN2O5. the western Mediterranean Sea around the Balearic MinimumNO columnsarepresentaftersunrisewhen 2 IslandsandtowardstheNorth-Africancoast(Fig.5, mostoftheNO isquicklyphotolyzedintoNO,thus 2 lowerleft). loweringthetotalcontentbyoneorderofmagnitude. Beforethecoldfrontreachedtheareaon8Novem- StratospherictotalNO levelsslightlyrecoverduring 2 ber2001thetypicalNO verticalcolumnwasaround day time, typically by about 1×1014 molecules 2 (2.5–3.5)×1015 ±3.5×1014 moleculescm−2,but cm−2 h−1. To evaluate the model performance we higheroverFrance,whileanunusuallyhighmaximum compare the simulated minimum NO values with 2 totalcontentof7.27×1015 ±4.4×1014 molecules available HALOE November observations in the cm−2wasobservedon11November2001.Anegative midlatitudes (Gordley et al., 1996) for sunrise con- biasofabout−1.4×1014moleculescm−2needtobe ditions (http://haloedata.larc.nasa.gov/Haloe/home. consideredtocorrectforpossibleartefactsinGOME html).AdirectcomparisonofthespatialNO distri- 2 backscatterdata(Martinetal.,2002).Thegivenerror butionswithindependentspace-bornemeasurements levelsarederivedfromtheslantcolumnfittingpreci- iscompromisedduetomissingdataduringtheperiod sion(i.e.standarddeviation),thecorrespondingcloud ofinterest.ROSENO minimumtotalcolumnvalues 2 fittingprecisionandanassumedprecisionofAMFsof above 20 km altitude compare well with HALOE 1%whichisduetoradiativetransfermodelling.Errors measurements with model values being slightly ofreferencespectrawerenotconsidered.Theorigin higher. Model results are in excellent agreement oftheNO plumeisfurtheranalysedinsection3.5. with HALOE observations near the stratospheric 2 Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 ALGERIANSEVEREWEATHEREVENT 1001 Fig.5. GOME-measurednitrogendioxidedistributionon10November2001(upperleft)and11November2001(lowerleft), andlightningstatisticsforbothonedayearlier,the9November2001(upperright)and10November2001(lowerright). SlightlyincreasedcolumndensitiesareobservedoverthestraitofGibraltarandalongtheCantabriccoaston10November 2001(upperleft),wherebothnight-timelightningactivitywasreported.On11November2001,unusuallyhighNO2column densitiesofabout7.2×1015moleculescm−2weremeasuredovertheMediterraneanSea(lowerleft).Lightningstatistics weretakenfromhttp://www.wetterzentrale.de. NO maximum around 30 km altitude. Modelled At the GOME overpass time during the period of 2 NO columns near the 100 hPa level (∼16 km) interest the temporal changes of NO due to chem- 2 2 tend to be higher than corresponding HALOE ical transformation are typically low. ROSE strato- values, but deviations stay within the known bias spheric NO columns calculated at 10:00 UTC de- 2 of HALOE. The simulated average zonal variability creasesmoothlyfrom2.1×1015moleculescm−2on of 10% for the NO column corresponds well with 6November2001to1.6×1015moleculescm−2on11 2 HALOEobservations. November2001at40◦N0◦E,whereGOME-derived Tellus55B(2003),5 teb301533/teb078 Tellus.cls October1,2003 14:29 1002 W.THOMASETAL. Fig.6. Totalstratosphericnitrogendioxidecontentfor10November2001(leftpanel)and11November2001(rightpanel) ascalculatedbyROSEatGOMElocaloverpasstimes(10:20LT).Themaximumstratosphericcontentisabout2.6×1015 moleculescm−2overtheSaharandesert,whileitisaround1.8×1015moleculescm−2intheareasouthwardoftheBalearic IslandswheremeasuredNO2valueswerehighest.Theday-to-dayvariationinthelatterarearemainssmall. NO levelswerehighest.Therefore,thetemporalres- November2001(seeFig.5,lowerleft)doesnotap- 2 olutionofUKMOanalysisdataofabout24hissuf- pearinthesimulations.Note,however,thatROSEdoes ficientlyhightodescribetheimpactofstratospheric notconsidertroposphericemissions(section2.2).The transport.Thevariationfrom10to12November2001 troposphericNO contentisfurtherdiscussedinthe 2 islowerthan5×1014moleculescm−2.ROSENO col- followingsection. 2 umnresultsatGOMEoverpasstime(10:20LT)on10 and11November2001areshowninFig.6. 3.5. Derivationandanalysisoftropospheric TheROSE-modelledspatialdistributionofNO is 2 nitrogendioxide dominated by the bulk of NO in the stratosphere 2 around30km.TherelativelylowNO columnsover Severalmethodstoderivethenitrogendioxidecon- 2 NorthernandCentralEuropeandevendowntoSpain tent in the troposphere are discussed in the litera- on10November2001canbeattributedtowave-like ture. Velders et al. (2001) apply both the so-called transport in stratospheric levels at higher latitudes. tropospheric excess method (Richter and Burrows Measured ozone fields and PV analysis indicate a 2002) and image processing technique to derive the morepronouncedmeridionaltransportinloweratmo- troposphericNO columnarcontent.Thetropospheric 2 sphericlevelsespeciallyon9November2001(Figs.3a excessmethodallowsonetoseparatetroposphericand andb).ThesimulatedNO columncontentduringthe stratospheric contributions to the total NO content 2 2 GOMEoverpasstimevariesbetween(0.9–2.6)×1015 under the assumption of a relatively stable (both in moleculescm−2fromNorthtoSouth,whichislower time and space) NO distribution over the free Pa- 2 thancorrespondingGOMEmeasurementsoutsidethe cificOcean.Laueretal.(2002)comparetropospheric NO plume over water surfaces and high latitudes. NO columnsderivedfromGOMEmeasurementsand 2 2 The strong increase of measured NO levels on 11 results of a coupled chemistry–climate model. The 2 Tellus55B(2003),5