Final Report MACC-III

148
Grant agreement n°633080 Final Report MACC-III Monitoring Atmospheric Composition and Climate 3 February 2016

Transcript of Final Report MACC-III

Page 1: Final Report MACC-III

Grantagreementn°633080

FinalReport

MACC-IIIMonitoringAtmosphericCompositionandClimate3

February2016

Page 2: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page2of148

Page 3: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page3of148

Date February2016

Status FinalVersion

Authors Seepage7-10

Use and reproduction of this report or parts of it may be restricted.Appropriatenon-commercialusewillnormallybegrantedundertheconditionthatreferenceismadetoMACC-III.Pleaseenquirebye-mailat:[email protected]

This document has been produced in the context of the MACC-III project (Monitoring AtmosphericComposition and Climate, phase 3). The research leading to these results has received funding from theEuropeanCommunity'sHorizon2020Programmeundergrantagreementn°633080.All information inthisdocument is provided "as is" and no guarantee or warranty is given that the information is fit for anyparticularpurpose.Theuserthereofusestheinformationatitssoleriskandliability.Fortheavoidanceofalldoubts,theEuropeanCommissionhasno liability inrespectofthisdocument,which ismerelyrepresentingtheauthorsview.

Page 4: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page4of148

Page 5: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page5of148

ExecutiveSummary/Abstract

TheMACC-III (Monitoring Atmospheric Composition and Climate – Phase III) project was the lastinterim stage in the development of the operational Copernicus Atmosphere Monitoring Service(CAMS): its overall objectivewas to function as the bridge between the developmental precursorprojects-GEMS,PROMOTE,MACCandMACC-II-andthestartofCopernicusoperations.

MACC-IIIwasfundedunderHorizon-2020andactivitiesranfromAugust2014toJune2015,afterano-costextensionwasagreedbytheResearchExecutiveAgency.Itscontinuedprovisionofcoherentatmospheric data and information, either directly or via value-adding downstream services, havebeen for the benefit of European citizens and helped meet global needs as a key Europeancontribution to the Global Climate Observing System (GCOS) and the encompassing Global EarthObservationSystemofSystems(GEOSS).Itsservicescovered:

• regionalairqualityandglobaltransportofatmosphericpollutants;• climateforcings;• stratosphericozoneandUVradiation;• solar-energyresources;• emissionsandsurfacefluxes.

The website of MACC-III (http://copernicus-atmosphere.eu initially, now available at:http://macc.copernicus-atmosphere.eu) gives access to the searchable catalogue of products andoutputsoftheproject.Foreach individualproduct, linkstoquick-lookplots,verification/validationresultsandtothecorrespondingnumericaldataareprovided.Thewebsitehasalsoareasprovidingbackgroundinformationontheprojectandontheservicesdelivered.Further,ithasa“news”andan“infocus”sections,whichhighlightedMACC-IIIresponseincaseofeventsofspecificinterest(e.g.airqualityepisodes,wildfires,volcaniceruptions…)orofprojecteventssuchasGeneralAssembliesorUserevents.Atotalof186deliverableshavebeendeliveredduringthecourseofMACC-III.Thevastmajorityoftheseareavailableat:

http://macc.copernicus-atmosphere.eu/documents/macciii/deliverables/

The present final report provides a synthetic account of activities during MACC-III and presentshighlights of themain results and findings. It is presented with chapters on each of the 15 sub-projects. A companion Special Issue in the open access European journals ACP/AMT/GMD/ESSDprovidesamorein-depthanalysisofsomeoftheresearchresults.Allthecorrespondingpaperscanbefreelyaccessedathttp://www.atmos-chem-phys.net/special_issue310.html.

Page 6: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page6of148

FouroftheAdvisoryBoardmembersgavetheirviewsontheachievementsoftheMACCseriesofprojectsduringtheMACC-IIIGeneralAssembly(January2015)andprovidedtheirinternationalperspective.

Page 7: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page7of148

Co-authorsandcontributorsManagement(MAN)ECMWF: Vincent-Henri Peuch (project co-ordinator), Richard Engelen (project manager), RebeccaCalnan(projectassistant)

BIRA-IASB:Jean-ChristopherLambert,AnnedeRudder

Acquisitionofobservations(OBS)NILU:LeonorTarrason(leader),PhilippSchneider,ÅsmundFahreVik,WilliamLahoz,MatthiasVogt

ECMWF:MartinSuttie,MarijanaCrepulja

NUIG:ColinO’Dowd,TomasGrigas

CNRS-LA:ValérieThouret,HannahClark

JULICH(collaboration):MartinSchultz,SnehalWaychal

Emissions(EMI)UPMC:ClaireGranier(leader),KaterinaSindelarova,ThiernoDoumbia

EC-DGJRC:GreetMaenhout-Janssens,MonicaCrippa,MarilenaMuntean

INERIS:FrederikMeleux,BertrandBessagnet

TNO:HugoDeniervanderGon,JeroenKuenen

METNorway:MichaelGauss

CNRS-LA(collaboration):CatherineLiousse,AudeMieville

ObservatoireMidi-Pyrénées(collaboration):SabineDarras

NOAAESL,USA(collaboration):GregoryFrost

NCAR,USA(collaboration):Jean-FrancoisLamarque

Fireemissions(FIR)ECMWF:JohannesKaiser(leader),SamuelRémy

JULICH:AngelikaHeil,MartinSchultz

IPMA:JoaoMacedo,IsabelTrigo

KCL:JiangpingHe,RonanPaugam,MartinWooster,WeidongXu

VUA:NielsAndela,RobDetmers,GuidovanderWerf

Greenhousegases(GHG)LSCE: Frédéric Chevallier (leader), Philippe Bousquet, Robin Locatelli, Anna Lourantou, JérômeTarniewicz,LynnHazan

Page 8: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page8of148

IUP-UB: Maximilian Reuter,Jens Heymann, Michael Hilker, Vladimir Rozanov, Alexei Rozanov,MichaelBuchwitz

LeicesterUniversity:HartmutBoesch,KristiinaByckling

CNRS-LMD:RaymondArmante,CorinneBurlaud,LaurentCrépeau,CyrilCrevoisier

ECMWF:AnnaAgusti-Panareda,SébastienMassart

JRC:MihaiAlexe,PeterBergamaschi

NILU:RonaThompson

TNO:ArjoSegers

SRON:RobDetmers,RemcoScheepmaker,OttoHasekamp,IlseAben

NASA/JPL(collaboration):ChristianFrankenberg

KIT(collaboration):AndreButz

Globalreactivegases(GRG)JULICH:MartinSchultz(leader),OlafStein,SabineSchröder,SnehalWaychal

ECMWF:JohannesFlemming,AntjeInness,LukeJones,BeatrizMonge-Sanz,MarkParrington

KNMI:VincentHuijnen,RolandvanderA,HenkEskes

BIRA-IASB:SimonChabrillat,KarolienLefever,DominiqueFonteyn

MF-CNRM:VirginieMarécal,JoaquimArteta,BéatriceJosse

CERFACS:DanielCariolle,EmanueleEmili,ElodieJaumouillé

IUP-UB:Anne-MarleneBlechschmidt,AndreasRichter,AndreasHilboll

UPMC:IdirBouarar

DLR:FrankBaier,DiegoLoyola

RIUUK:HendrikElbern,KetevanKasradze,ScarletStadtler

Globalaerosols(AER)CNRS-LMD: Olivier Boucher (leader),NicolasHuneeus, Jeronimo Escribano, AlinaGainusa-Bogdan,CyrilCrevoisier,VirginieCapelle,AlainChédin

CNRS-ICARE: Jacques Descloitres, Anne Vermeulen, Nicolas Henriot, Stanislaw Matusiak,ManuelSaunier,JulienBonte,BrunoSix,SylvainNeut,Jean-MarcNicolas,LoredanaFocsa,HenriMeurdesoif

CEA-LSCE:FrédéricChevallier

DLR:ThomasHolzer-Popp, MiriamKosmale,DmytroMartynenko,FranziskaSchnell

ECMWF: Angela Benedetti, Alessio Bozzo, Luke Jones, Johannes Kaiser, Jean-Jacques Morcrette,SamuelRémy

FMI:GerritdeLeeuw,PekkaKolmonen,LarisaSogacheva

METNorway:MichaelSchulz,JanGriesfeller,AnnaBenedictow

MPIM:StefanKinne

Page 9: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page9of148

University of Leeds: Graham Mann, Will Hewson, Matt Woodhouse, Francois Benduhn, SandipDhomse

UniversityofLeipzig:JohannesQuaas,JohannesMuelmenstaedt,KarolineBlock

UniversityofReading:NicolasBellouin,AnnaEsteve

GlobalProduction(GDA)ECMWF:RichardEngelen(leader),AnnaAgusti-Panareda,AngelaBenedetti,AlessioBozzo,MarijanaCrepulja, Johannes Flemming, Jan Haseler, Antje Inness, Luke Jones, Johannes Kaiser, SébastienMassart, Beatriz Monge-Sanz, Jean-Jacques Morcrette, Mark Parrington, Miha Razinger, SamuelRémy,MartinSuttie,XiaoboYang

Validationactivities(VAL)KNMI:HenkEskes(leader),VincentHuijnen,RonaldvanderA

AA:TheodoraAntonakaki,JohnKapsomenakis,MihalisVrekoussis,ChristosZerefos

AEMET:EmilioCuevas

AUTH:DimitrisMelas,EleniKatragkou

IUP-UB:Anne-MarleneBlechschmidt,AndreasRichter

BIRA-IASB:E.Botek,SimonChabrillat,Y.Christophe,B.Langerok,KarolienLefever

CNRS-LA:HannahClark,AudreyGaudel,ValérieThouret

CNRS-LMD:NicholasHuneeus

DWD:HaraldFlentje,WernerThomas,AnnetteWagner

ECMWF:LukeJones,MihaRazinger

FMI:AnttiArola

METNorway:AnnaBenedictow,JanGriesfeller,MichaelSchulz

MPG:StefanKinne

UPMC:IdirBouarar

RegionalAirQualityactivities(EDA,ENS,EVA)RIUUK: Hendrik Elbern (leader EDA), Elmar Friese, Ketevan Kasradze, Zoi Paschalidi, Kai Krajsek,ClarissaFigura,ScarletStadtler

AEMET:AlbertoCansado,IsabelMartínez,TomásMorales

AUTH: Dimitrios Melas, Stavros Avgoloupis, Spyros Dimopoulos, Theodore Giannaros, ChristosGiannaros,EleniKatragkou,CharikliaMeleti,NatasaPoupkou

CERFACS:DanielCariolle,EmanueleEmili,ElodieJaumouillé

CNRS-LISA:MatthiasBeekmann,AdrianaComan,GillesForet,BenjaminGaubert

FMI: Mikhail Sofiev, Julius Vira, Marje Prank, Joana Soares, Rostislav Kouznetsov, Ari Karppinen,JaakkoKukkonen

Page 10: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page10of148

INERIS:LaurenceRouïl(leaderEVA),BertrandBessagnet,FrédérikMeleux,AnthonyUng

KNMI:HenkEskes,UjjwalKumar,MartijndeRuyterdeWildt,RobertvanVersendaal

METNorway:M.Gauss,AlvaroValdebenito,AnnaBenedictow,HeikoKlein,BirtheMarieSteensen,SvetlanaTsyro,PeterWind

MF-CNRM: Virginie Marécal (leader ENS), Joaquim Arteta, Nicole Asencio, Guillaume Beffrey,Frederic Chavaux, Françoise Chéroux, Jean Clochard, Richard Dupont, Laaziz El Amraoui, DenisFerriol,ChantalFlick,SylvieGuidotti,JonathanGuth,MathieuJoly,BéatriceJosse,NikolayKadygrov,AntoineKergomard,BrunoLacroix,VincentLemaire,StéphaneMartinez,PhilippeMoinat,JonathanParmetier,Dominique Paulais,Marion Pithon,Matthieu Plu, SolenQuéguiner, SébastienRouzeau,BojanSic,PascalSimon

NILU:LeonorTarrason

SMHI: LennartRobertson,CamillaAndersson,StefanAndersson,ManuAnnaThomas,CeciliaBennet,RobertBergström,MagnuzEngardt,MichaelKahnert

TNO:LyanaCurier,MartijnSchaap,ArjoSegers,RichardKranenburg,RenskeTimmermans

ProductsinsupportofPolicyusers(POL)METNorway:MichaelGauss(leader),SemeenaValiyaveetil,AlvaroValdebenito

NILU:LeonorTarrason,SamErikWalker

EAA:HerbertHaubold,ChristianAnsorge,ChristianNagl

CNRS-LISA:IsabelleColl,GillesForet,GuillaumeSiour,MatthiasBeekmann

INERIS:LaurenceRouil,FrederikMeleux

Solarradiation(RAD)DLR:MarionSchroedter-Homscheidt(leader),NielsKillius,GerhardGesell

Armines:LucienWald,BellaEspinar,MireilleLefèvre,PhilippeBlanc

FMI:AnttiArola,MikkoPitkänen,VaidaCesnulyte,AndersV.Lindfors

ECMWF:AngelaBenedetti,AlessioBozzo

Transvalor(collaboration):EtienneWey

UserInterface(INT)DLR: Thomas Holzer-Popp (leader), Miriam Kosmale, Dmytro Martynenko, Lars Klüser, FranziskaSchnell,JulianMeyer-Arnek,OlegGoussev,ChristophHarsch,ThiloErbertseder,GerhardGesell

ECMWF:RichardEngelen,XiaoboYang,MihaRazinger

CERC:AmyStidworthy,DavidCarruthers

JULICH:MartinSchultz,SnehalWaychal

UPMC:ClaireCranier,KaterinaSindelarova,ThiernoDoumbia,IdirBouarar

CNRS-LA(collaboration):CatherineLiousse

Page 11: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page11of148

Tableofcontents

1.Projectobjectivesandmanagement(MAN)..............................................................................14

2.AcquisitionofObservations(OBS)............................................................................................19

3.Emissions(EMI)........................................................................................................................31

4.FireDataAssimilation(FIR).......................................................................................................41

5.Greenhousegases(GHG)..........................................................................................................53

6.GRG:GlobalReactiveGases......................................................................................................63

Page 12: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page12of148

7.GlobalAerosols(AER)...............................................................................................................73

8.Globalintegrateddataassimilation,productionandservices(GDA)........................................79

9.Validationactivities(VAL).........................................................................................................87

10.DataassimilationforEuropeanairquality(EDA)....................................................................99

11.Europeanensembleair-qualityanalysesandforecasts(ENS)...............................................111

12.Validatedairqualityassessment(EVA)................................................................................117

13.Solarradiationservices(RAD)...............................................................................................123

Page 13: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page13of148

14.Regionalpolicysupport(POL)...............................................................................................137

15.UserInterfaceactivities(INT)................................................................................................143

Page 14: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page14of148

1.Projectobjectivesandmanagement(MAN)

Some of today’s most important environmental concerns relate to the composition of theatmosphere.Theincreasingconcentrationofthegreenhousegasesandthecoolingeffectofaerosolareprominentdriversofachangingclimate,buttheextentoftheirimpactisoftenstilluncertain.

AttheEarth’ssurface,aerosols,ozoneandotherreactivegasessuchasnitrogendioxidedeterminethe quality of the air around us, affecting human health and life expectancy, the health ofecosystems and the fabric of the built environment. Ozone distributions in the stratosphereinfluencetheamountofultraviolet radiationreachingthesurface.Dust, sand,smokeandvolcanicaerosols affect the safe operation of transport systems and the availability of power from solargeneration,theformationofcloudsandrainfall,andtheremotesensingbysatelliteofland,oceanandatmosphere.

Toaddresstheseenvironmentalconcernsthereisaneedfordataandprocessedinformation.

InthecontextoftheCopernicusprogramme(http://www.copernicus.eu),theMACC-IIIprojecthadtheoverallfunctionalobjectiveofdeliveringreliableoperationalproductsandinformationservicesthat support research, European environmental policy and the on-going development of user-specific downstream services. It prepared for the transition to long-term sustainable operation asthe fully-fledgedCopernicusAtmosphereMonitoringService (CAMS) fromthesecondhalfof2015onwards.

IntrackfromthepredecessorprojectsGEMS,MACCandMACC-III,theserviceswereoperatedandfurtherdevelopedinawaycomplementarytotheestablishedrangeofmeteorologicalandrelatedservices that are operated nationally and at European level by what is known collectively as theEuropean Meteorological Infrastructure. The latter services include most notably weatherforecasting,butalsoincludeservicesmorecloselyrelatedtotheCopernicusservices,suchasthoseprovided to international aviation by the Volcanic Ash Advisory Centres (VAACs). The stronginvolvement of meteorological service providers in the MACC-III consortium ensures that theCopernicusservicescanbenefitmost fully fromreadyaccesstothemeteorologicaldataanddata-processing infrastructure essential for their operation, and that the Copernicus services areimplemented in a way that supports the established services in a manner consistent with theEuropeanUnion’sprinciplesofcomplementarityandsubsidiarity.

TheMACC-IIIprojecthasstartedonAugust1st2014,withaninitialdurationof8months.InordertofacilitateservicecontinuitywithCAMS,activities intheprojectwere intheendextendeduntil theendofJune2015(andformalendoftheprojectuntiltheendofAugust2015).

1.1MACC-IIIProducts

Consistent with the pre-existing capabilities and service lines developed as part of the series ofprecursor projects, requirements expressed in the H2020 Call, and requirements determinedthroughuserconsultation,MACC-IIIsuppliedcontinually:

• monitoringoftheglobaldistributionsofgreenhousegases,reactivegasesandaerosolsthroughassimilationofsatelliteandinsituobservations,usingNRT,delayed-modeandreanalysisproductionsystems;

Page 15: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page15of148

• twice-dailyforecastsoftheglobaldistributionsofreactivegasesandaerosolsforseveraldaysahead;

• specificstratosphericozoneproducts,basedbothontheintegratedMACC-IIIsystemandonsystemsthatareoperatedtoextendthelong-termrecordsbuiltupinPROMOTEandthenMACC/MACC-II;

• boundaryvaluesforregionalmodellingoftroposphericandstratosphericchemistry,andlocalandurbanmodellingforairquality;

• analysesandforecastsfortheEuropeandomainbasedonanensembleapproachusingmultipleregionalairqualitymodels;

• annualassessmentsandsourceattributionforthemainatmosphericpollutantsoverEurope;• toolsthatmaybeappliedtopastcasesorinNRTtoassessactionstocontrolpollution

events;• globalfireanalysesandestimatesofemissionsfromfiresforuseintheglobalandEuropean

regionalmonitoringandforecastingsystems;• surfacefluxesofcarbondioxide,methaneandaerosolsproducedusinginversemethods;• globaldatasetsforemissionsfromsourcesotherthanfires,tobeupdatedbasedonnew

statisticsorresultsfromfluxinversion;• higher-resolutionemissiondatasetsforaerosolsandreactivegasesoverEurope;• satellitedataretrievalsasneededtocomplementworkcarriedoutunderspace-agency

auspices,includingthatfromtheEUMETSATSAFsandESAactivitiessuchasitsClimateChangeInitiative(CCI);

• estimatesofdirectandindirectclimateforcingfromaerosols;• coredataservicessupportingsolarpowergenerationandmonitoringandpredictionofUV

radiation.

MACC-III’sproductsarefreelyandopenlyavailabletodownstream-serviceprovidersandotherusersthroughoutEurope.MACC-IIIanditsdownstreamservicesectorwillbetweenthemenableEuropeancitizensathomeandabroad tobenefit from improvedwarning, advisoryandgeneral informationservices and from improved formulation and implementation of regulatory policy. MACC-III,together with its scientific-user sector, also helps to improve the provision of science-basedinformationforpolicy-makersandfordecision-makingatalllevels.

1.2Projectco-ordination

MACC-IIIhasdelivereddataandserviceswithahigh-levelofreliabilityanddemonstratedeffectivereadiness for the transition to the fully operational Copernicus Atmosphere Monitoring Service(CAMS).

ApracticaldifficultyforMACC-IIIwastoreconciletherampingupactivitiesofCAMSontheonehand-with ECMWF formally entrusted with the implementation tasks of the operational service (aDelegationAgreementwas signedwith the EuropeanCommissionon 11/11/2014, see Figure 1.1)andsettingupcompetitiveInvitationsToTenderfortheprovisionofthedifferentserviceelementsgoing forward and according to the intangible rules of the European Union (openness, fairness,transparency)- and, on the other hand, the management by ECMWF of the consortium of theprovidersofMACC-IIIpre-operationalservices.Asduringthe lastphaseofthepredecessorprojectMACC-II, the focus inMACC-IIIwas in testing andupdating the documentation of key aspects foroperationalimplementationratherthanoneffectivelygoingforwardwiththeeffectivetransition–inordernottodistortthecompetitiveprocessgoingoninparallel.ECMWFhasalsomadesurenotto

Page 16: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page16of148

disclose privileged information to MACC-III partners about developments in CAMS, but alwayscommunicatedpublicly(e.g.viaan“OpenDay”onupcomingITTsinBrusselsinFebruary2015).

Figure1.1.SignatureoftheDelegationAgreementbetweentheEuropeanCommission(DG-GROW)andECMWFfortheimplementationoftheCopernicusAtmosphereMonitoringService(CAMS)andoftheCopernicusClimateChangeService(C3S).

A large fraction of themanagement and co-ordination activitieswas in liaisingwith other relatedprojectsandprogrammes,includingCopernicusdownstreamanduseruptakeprojects,aswellasin-situ data coordination (European EnvironmentAgency). Regarding theCopernicus space segment,ECMWFparticipatedasamembertotheMissionAdvisoryGroupsofSentinel4-5andofSentinel5Pandismeetingbi-annuallywithEUMETSAT.

In direct track fromMACC-II, communications activities inMACC-III have remained at a high levelandproject representation indifferentconferencesandeventshasbeenavery significantactivityforthecoordinatorandprojectmanagerinparticular.

The “MACC” Special Issue in the European open-access journals ACP/AMT/GMD/ESSD has beenopenuntil the endof 2014 and itwas a great success in termsof thenumber andquality of thesubmittedmanuscripts.Abookwiththepapersfromthewillbeprintedin2016,whichwillserveasalegacydocumentfortheseriesofMACC,MACC-IIandMACC-IIIprojects.

ThescheduledreviewsforMACC-IIItookplaceasplanned:theGeneralAssemblytookplaceon19-21January2015atECMWFinReading(UK),andafinalreviewtookplaceonJuly15th2015atREAinBrussels. The project benefited from the input provided by the project Advisory Board, whoattended the third General Assembly. The AB provided positive and useful feedback in mostscientific, technical and operational areas of the project, linking also with satellite and in-situobservationsaswellaswithsomeimportantsegmentsofusers(policy,research…).

Page 17: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page17of148

Figure1.2.Morethan100personsparticipatedtoMACC-IIIGeneralAssemblyatECMWFinReading(19-21January2015).

Asignificantmanagementeffortwastoprepareandimplementanamendment inordertoextendcertainactivitiesbeyondtheinitialendofMACC-IIIinMarch.ThiswasachievedbymovingoutmostcostsofECMWFinMACC-IIIontoCAMSandre-distributingtheequivalentgrantstothepartnersandactivities that needed to be maintained most from a service continuity point of view: productsdelivery,supporttousers...

1.2Servicespecificationsandtransitiontooperations

Thehigh-levelobjectiveswereto:

• update continually the products and service specifications document and to monitorcompliancewithuserrequirementsandwithbestinternationalstandards;

• facilitatethetransitiontotheCopernicusAtmosphereMonitoringService,bydocumentingoperationalproceduresastheyreachmaturity.

1.2.1ServiceSpecifications

Product and service specifications are needed to informpotential users of the product or servicesuitability to some given range of uses. A priori, specifications are also needed as a reference tomeasure progress towards service targets and continuity of quality level. A final update of the“ServiceSpecificationDocument”(SSD,D68.1)hasbeenreleasedduringMACC-III.Thisdocumentisfully in sync with the online catalogue, as well as with the products portfolio (effectively ashort/syntheticversionoftheSSD).Asingle/centralproductsspecificationsmetadatadatabasehasbeen established in order to ensure consistency across the different information provided on theproducts. This tool has proven very effective and will be maintained in the operations phase. InCAMS,thisdocumentwillcontinuetoberegularlyupdatedasaresultoftheinteractionswithusersand as technical/operational capabilities ramp up, resulting in (among other aspects) improvedreliabilityandtimelinessspecifications.

1.2.2TransitiontothefullyoperationalAtmosphereService

Anumberofpre-operationalprocedureshavebeenputinplaceinthepredecessorprojectsMACCandMACC-II(e.g.decisionproceduresingeneral,cycleupgrades,validationprocedures,automationprocedures, collection of user feedback, user support, user information…). While most of theinformation is available online on the MACC-III website, a specific deliverable (D68.2, “Outline

Page 18: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page18of148

operational procedures for the Copernicus Atmosphere Service”) describes the high-levelprocedures for the global production streams operated at ECMWF at the end of MACC-III. TheoperationalproceduresfortheothermainserviceelementofMACC-III,theregionalproduction,aredescribedinanotherdeliverablereport(D52.1,“Updatedreportontheoperationalimplementationincludinganalysisofopportunitiesforoperationalservicecertification”).

Figure1.3comparestheMACC-IIIpre-operationalglobalproductionrunatECMWFintheformofaGantt chart describing the daily sequence of tasks (top) with the situation foreseen for theoperational configuration (bottom), as it will be implemented during the Copernicus AtmosphereMonitoringservice.Themainproductionsequencecurrentlyhasadurationofabout8hours,anditwill shrink to justover3hours.Asa result, thebase timedaily for the forecastswillbe12hnooninsteadof0h:theanalysesandforecastswill thusbenefit fromfreshersatelliteobservations,withqualitativeimprovementsexpected.

Figure1.3.GanttchartofthecurrentMACC-IIInear-real-timeproductionsystem(top)andoftheenvisagedCAMSoperationalnear-real-timeproductionsystem(bottom)showingthevarioustasksinvolvedasafunctionoftimeinUTC.

Another important operational aspect is how system upgrades are implemented. This aspect isalreadyfullyfunctional.Briefly,whenanewcandidateconfigurationisready,itisrunforaperiodofafewmonths(e-suite)inparalleltotheoperationalproduction(o-suite).Theperformanceofthee-suiteisevaluatedagainsttheo-suiteaswellasindependentobservations.Iftheperformanceofthee-suite is deemed satisfactory, after giving users a period of time to adjust (at least a month ingeneral), the e-suite eventually replaces the o-suite. Such procedure, which mimics the oneemployedforchanges inoperationalNumericalWeatherPrediction,ensuresnon-regressionofthesystem’s performance and provides users with time to accommodate to the successive newconfigurations

Page 19: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page19of148

2.AcquisitionofObservations(OBS)

ThepurposeofOBS sub-projectofMACC-III is to link toexistingobservational infrastructuresanddata.Thegoal is toacquire theobservationaldataneeded foruse in theproductionsystems.TheOBSsub-projectdrawsonexistingobservationcapacitiesandworkstoensuretheestablishmentofa(pre-)operationaldataflowbetweendataprovidersandtheMACC-IIIproductionchain.

Inparticular,theOBSsub-projectworked:

• toidentifyexistingdatabasesandobservationalinfrastructure,byevaluatingtherelevanceoftheexistingdataflowsforoperationalMACC-IIuseandtheirpossiblecompliancewithINSPIREandWISstandards;

• to operationalize the data transmission and data exchange, by linking to existing databasesandsecuringaccesstodatafortheglobalandregionalproductionsystems;

• toevaluatetheuseofnewsatelliteproductsastheybecomeavailable,bycarryingoutsomeinitialtestingofthesatelliteproductsagainstexistingNRTinsituobservations;

• topreparefeedbacktotheobservationdataproviders,bydrawingontheexperiencegainedfromdatauseinMACC-II’sglobalandregionaldataassimilationsystems.

InMACC-III, theOBS sub-projecthasbuilton the (pre-)operationaldataacquisitioneffort carriedout onMACC-II. This is bothwith respect to access toNRT in situand satellite data necessary tosupporttheMACC-IIIproductionchainandwithrespecttoaccesstodelayed-modeandreprocessedversions of data originally in near-real time acquired for use in MACC-III’s delayed-mode andreanalysis production streams, as well as for validation purposes. The acquisition data flowsimplementedand tested inMACC-II haveproceededwithoutproblems inMACC-III. In addition tpexisting data flows, the availability of aerosol research data from different programs not yetincorporatedintheproductionchainhasbeenidentifiedaspartofMACC-III/OBS.

Threemain reportshavebeenproduced inOBS focusingonvalidationof satelliteproductsduringMACC-III.TheseincludeareportonthevalidationoflidarCALIOPdataversusEARLINETobservations(D15.1)andareportonthecapabilitiesofusingMAX-DOASdataforsatelliteproductvalidation,inparticular TROPOMI NO2 data (D15.2). The third report (D16.1) introduced the methodology ofObservingSystemSimulationExperiments(OSSEs)asawayofquantifyingtheadded-valueoffuturesatellitemissionsandrecommendationsareprovidedastohowMACC/CAMSmodellingsystemscancontributetoOSSEsforconstituentsaffectingairquality.

ThemaindeviationfromtheplannedworkofOBSinMACC-IIIhasbeentheaccesstovalidatedandNRTdata though thenewEEAe-reporting system. The regulatory requirement to report 2013 airquality data via the new e-reporting system has been challenging formanyMember States. As aresult, the validated surface air qualitydatawas first available toMACC-III partners at theendofJune 2015. This delivery has imposed unavoidable delays to the regional production chain forregionalreanalysis(EVA).Itshouldbenoted,however,thatthereasonsforthedelaylaybeyondthemandateandscopeoftheMACC-IIIteams.

2.1Operationalsatelliteproducts

Themainobjectivesoftheseactivitiesweretoensuretheroutineprovisionofsatelliteobservationsfor the NRT system for analysis and prediction of atmospheric composition and for the delayed-

Page 20: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page20of148

mode analysis. Theworkwas organized in three different tasks that have progressed as plannedduringMACC-III.

2.1.1Acquisitionandpre-processingofsatelliteobservationsinNRT

ECMWFcontinued tomaintainand furtherenhance the suiteofobservationaldata streams, fromboth operational and research satellite missions, used in the MACC-III global monitoring andforecasting system.An overviewof the existing satellite observations used for the global and theregionalproductionchainsisprovidedinTable2.1andTable2.2below.

Table2.1.SatelliteobservationscurrentlyassimilatedintheglobalNRTMACC-IIIproductionsystem

Instrum. Satellite SpaceAgency DataProvider Species Status*MODIS EOS-Aqua,-Terra NASA NASA AOD,FRP AMLS EOS-Aura NASA O3profile AOMI EOS-Aura NASA KNMI O3,NO2,SO2 ASBUV-2 NOAA-16,17,18,19 NOAA NOAA O3profile AIASI METOP-A EUMETSAT/CNES ULB/LATMOS CO AMOPITT EOS-Terra NASA NCAR CO AGOME-2 METOP-A,-B EUMETSAT/ESA DLR O3 AGOME-2 METOP-A,-B EUMETSAT/ESA DLR NO2,SO2 MIASI METOP-B EUMETSAT/CNES ULB/LATMOS CO MSEVIRI METEOSAT EUMETSAT LandSAF O3,FRP MImager GOES-11,-12 NOAA UCAR FRPradiances MCALIOP CALIPSO NASA lidarbackscat. POMPS SuomiNPP NASA O3 PIASI METOP-A,-B EUMETSAT/CNES EUMETSAT O3radiances PImager MTSAT-2 JMA JMA FRPradiances PVIIRS SuomiNPP NASA AOD,FRP PSEVIRI MSG EUMETSAT ICARE AOD P

*A=Activeassimilation;M=passivemonitoring;P=implementationisplanned.

Table2.2.SatelliteobservationscurrentlyassimilatedintheregionalNRTMACC-IIIproductionsystem

Instrum. Satellite SpaceAgency DataProvider SpeciesMODIS/SEVERI

EOS-Terra NASA NASA AODOMI EOS-Aura NASA KNMI NO2

GOME-2 METOP-A-B EUMETSAT/ESA DLR NO2IASI METOP-A EUMETSAT/CNES ULB/LATMOS O3profileMOPITT EOS-Terra NASA NCAR CO

2.1.2Acquisitionandpre-processingofhistoricalsatelliteobservations

ECMWFcontinuedtoacquireandstoreobservationsforreanalysisanddelayed-productionrunsaswell as to maintain overviews of the satellite data used. The overviews are available athttp://macc.copernicus-atmosphere.eu/cams-input-data

Page 21: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page21of148

2.1.3Acquisitionandpre-processingofNO2datafromOMItoMACC-III

ThistaskwastosecurethatthedailydeliveryofNRTNO2datafromOMItoECMWFwascontinuallyguaranteed,securingthatupgradesoftheKNMIprocessory(e.g.relatedtoinstrumentchanges)takeplaceregularlyandthatthereisaquickresponseproblemfixingifdeliveryfails.

2.2In-situdataacquisition

The objective of this work package was to ensure the routine provision of ground-basedobservations for the operational forecast global and regional production and the provision ofvalidatedobservationsforthedelayed-modeanalysissystem.

2.2.1Acquisitionandpre-processingofin-situobservationsinNRT

Surfaceobservationsofatmospheric componentscontinued tobeacquired inNRT fromthesamenetworksidentifiedandtestedinthecourseofMACCandMACC-II.ThedataflowofNRTEuropeanairqualitydatafromEEAisnowoperationalisedatMeteoFrancebasedontheretrievalalgorithmsdevelopedinMACC-II.SomesurfaceobservationsfromtheWMOGAWaremadeavailableinnear-real-timebyDWDandareacquiredforuseintheverificationoftheglobalsystem.Inaddition,theglobal validation system uses aerosol characterization data such as aerosol light backscatteringcoefficient, aerosol light scattering coefficient, aerosol absorption coefficient from thenephelometerandphotometerdataatGAWWorldDataCenterforAerosols,hostedatNILU. Theacquisitionof surfaceobservations from theNorthAmericanAIRNowservice continuedaswell asthe acquisition of ICOS surface data for CO2 and CH4. Table 2.3 summarises the main dataacquisitionsourcesforNRTsurfaceobservationsInMACC-III.

Table2.3.NRTground-basedobservationscurrentlyusedintheMACC-IIIproductionsystem

Network/source Observationtype

Provider Species

Europeanairquality surface EEA O3, CO, NO2, NOx, SO2, PM2.5, PM10 hourlyconcentrations

GAW surface DWD/NILU O3, CO, SO2, NO, NO2, PM10 hourlyconcentrations/aerosolcharacterizationdata

AIRNow surface USEPA O3, CO, SO2, NO, NO2, PM2.5, PM10 hourlyconcentrations

ICOS surface ICOSconsortium CO2,CH4

2.2.2Acquisitionandpre-processingofvalidatedin-situobservations

Thevalidatedground-basedobservationsusedinMACC-III includethenetworkslistedinTable2.4.AERONET provides globally distributed observations of spectral aerosol optical depth (AOD),inversionproducts,andprecipitablewaterindiverseaerosolregimes.ThesedatawereincorporatedtoMACC-II aswell as data from theNASAMicroPulse LidarNetwork (MPLNET) and continued inMACC-III. In addition to these data, MACC-III compiled also validated data from the above-mentionedNRTnetworksandthevoluntarycontributionfromtheEuropeanAeroallergenNetwork(EAN) with pollen data to allow the verification of the experimental regional ensemble pollenforecasts.

Page 22: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page22of148

Table2.4.Validatedground-basedobservationscurrentlyusedintheMACC-IIIproductionsystem

Network/source Observationtype

Provider Species

AERONET surface NASA aerosolopticalthickness,dailyvalues

Europeanairquality surface EEA O3, CO, NO2, NOx, SO2, PM2.5, PM10 validatedconcentrations

GAW surface DWD/NILU O3, CO, SO2, NO, NO2, PM10 validatedconcentrations/aerosolcharacterizationdata

AIRNow surface USEPA O3, CO, SO2, NO, NO2, PM2.5, PM10 validatedconcentrations

MPLNET surfacelidar NASA aerosollidarbackscatterEuropean AeroallergenNetwork surface EAN Pollencharacteristicsforvalidation

ICOS surface ICOSconsortium CO2,CH4

2.2.3Identifynewin-situobservationdata

Aerosols are a key component in atmospheric composition. The aerosol research surface datacurrently in the validation and production chain ofMACC-III is a subset of the existing data fromdifferentnetworksattheEBASdatanodehostedatNILU.Thecurrentlyusedsurface/ground-baseddataislargelyconcurrentwiththeinformationavailableatWMOGlobalAtmosphericWatchWorldDataCentreforaerosol (GAW-WDCA).As indicated inTable2.5, linkingtoGAW/WDCAprogramishighlyadequateconcerningaerosolphysicalproperties.However,whenthevalidationdemandsofthe CAMS include the need for further chemical aerosol specification, additional data acquisitionefforts should be addressed to incorporate also data (like OC/EC data) from the InteragencyMonitoringofProtectedVisualEnvironments(IMPROVE)atUSandtheEMEPprograminEurope. Table2.5.Availabilityofseveralaerosolcharacterizationparametersfromdifferentresearchnetworks.Notethatthenumberofstationsavailablevariesfromnetworktonetworkandthatthetotalnumberofstationsisnotthesumfromthedifferentnetworks.Thisisbecausethesamestationcanreportdatatodifferentresearchnetworks.

AEROSOLRESEARCHDATA VALIDATEDQA/QCdata NRTdata

Parameter Program #Stations Program #Stations

aerosol_light_scattering_coefficient,aerosol_light_backscattering_coefficient

91 23

GAW-WDCA 89 GAW-WDCA_NRT

22

IMPROVE 41 NOAA-ESRL_NRT

18

ACTRIS 24 EUSAAR_NRT 3 NOAA-ESRL 21 NILU_NRT 1 EMEP 17 EUSAAR 17 NILU 3 CREATE 3 GUAN 1 EUCAARI 1 aerosol_absorption_coefficient 52 14

GAW-WDCA 50 GAW-WDCA_NRT

13

EUSAAR 20 NOAA- 12

Page 23: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page23of148

ESRL_NRT NOAA-ESRL 20 EUSAAR_NRT 3 ACTRIS 19 NILU_NRT 1 EMEP 18 GUAN 9 NILU 5 EUCAARI 4 CREATE 2 aerosol_optical_depth 24 18

GAW-WDCA 24 GAW-WDCA_NRT

18

NILU 1 particle_number_size_distribution 42 2

GAW-WDCA 35 EUSAAR_NRT 2 EMEP 23 NILU_NRT 1 EUSAAR 23 GAW-

WDCA_NRT1

ACTRIS 19 GUAN 14 CREATE 9 EUCAARI 6 NILU 5 particle_number_concentration 34 0

GAW-WDCA 32 NOAA-ESRL 18 EUSAAR 11 EMEP 10 ACTRIS 8 CREATE 3 NILU 1 equivalent_black_carbon(online,filter_absorption_photometer)

23 0

GAW-WDCA 17 EMEP 9 EUSAAR 9 ACTRIS 6 NOAA-ESRL 4 NILU 1 CREATE 1 EUCAARI 1 CAMPAIGN 1 EC,OC,TC(filter) 236 0

IMPROVE 186 EMEP 37 CAMPAIGN 20 GAW-WDCA 16 EUSAAR 14 ACTRIS 8 NILU 5 EUCAARI 4 AMAP 1 CREATE 1 EC,OC,TC(online) 2 0

Page 24: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page24of148

EMEP 2 CAMPAIGN 2

ThemaindeviationfromtheplannedworkofOBSinMACC-IIIhasbeentheaccesstovalidatedandNRTdata though thenewEEAe-reporting system. The regulatory requirement to report 2013 airquality data via the new e-reporting system has been challenging formanyMember States. As aresult, the validated surface air qualitydatawas first available toMACC-III partners at theendofJune 2015. This delivery 3 months after the end of the MACC-III project period has imposedunavoidabledelaystotheregionalproductionchain.Itshouldbenoted,however,thatthereasonsfor thedelay laybeyond themandateand scopeof theMACC-III teams. Service level agreementswithEEAinthecontextoftheCopernicusAtmosphereMonitoringServiceshouldincludecorrectivemeasurestominimizetheimpactfromthistypeofsituations.

2.3Aircraftmeasurements

The objective of this work package was to secure the timely access of aircraft observations ofatmosphericcompositionfromcommercialaircraft.TheworkinMACC-IIIwascarriedoutbyECMWFthatcontinuedtheeffortsofCNRS/LAlinkingtotheIAGOSResearchInfrastructure.

2.3.1Acquisitionandpre-processingofIAGOSaircraftobservationsinNRT

ThemostimportantIAGOSproductsforMACC-IIIincludeaircraftinsitumeasurementsof:O3,NOx,CO2,CH4 andaerosols. ECMWF isnow responsibleof checking thenear-real-timedata streams toensureresilienceandreliabilityandprepareoverviewsofdataexchangetofacilitatecommunicationwithGDAS,EVAandENSsubprojects.

AtpresentonlyozoneandCOdataareoperationallydeliveredtoECMWF,andfurtherdistributedtoMACC-III relevant partners. IAGOS data have been delivered in NRT (data are transmitted afterlanding) toMACC-II since July 2011. Ozone and CO profiles are automatically delivered in BUFRformat for testing assimilation procedures. The data aremade available at the CNRS/LA server inToulouse and ECMWFhas automatic and regular requests to check if new files are available. Theozone and CO NRT profiles (up to 12 km) are used to validate the model forecasts and fullyvalidated/calibrated data are used 6-12 months later for validating reanalysis (seehttp://www.iagos.fr/macc fordetails).Auniquecharacteristicof IAGOS is toprovidesimultaneousmeasurementsofozoneandprecursors in thecriticalUTLS region.ModelsdataaresystematicallyextractedalongtheflightpathsforevaluationasillustratedinFigure2.1.

Page 25: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page25of148

Figure2.1.ComparisonofOzoneandCarbonMonoxideprofilesfromIAGOSoverFrankfurtwithmodelleddatafromMACC.

IAGOSdatadelivery in real timewasalso facilitatedby the IGASFP7EUproject (http://www.igas-project.org,coordinatedbyC.Gerbig).DuringtheCopernicusAtmosphereMonitoringService, it isenvisaged to initiate the near-real access to measurements of aerosols, nitrogen oxides andgreenhousegases (CO2andCH4) from IAGOSasof ascentanddescentprofilesand cruisealtitudedata.

2.4Verticalprofileandcolumndata

Themainobjectiveofthisworkpackagewasto liaisewithdataprovidersofground-basedremotesensingdatainNRTandindelayedmode.Researchcolumnandprofiledataarecurrentlyavailablethroughalargenumberofnetworksandprograms,includingNDACC,AERONET,EARLINET,MPLNET,TCCONandSKYNET.The levelofoperationalizationoftheobservationaldataflowexchangevariesfor the different networks and it is currently done in an ad-hoc basis. Experience within theoperationalizationofobservationaldataflowsinMACC/OBSindicatesthatsupportingprojects(likeIGASforIAGOS,ICOS-INWIREforICOSandNORSforNDACC)arenecessarytobridgetheneedsforoperations at the Copernicus production service. This is particularly the case with ACTRIS whereadditionalsupportingagreementswillbenecessarytoallowforthedevelopmentofthealgorithmsneededforthetransmissionofground-basedaerosoldatafromEARLINETtoMACC/CAMS.Servicelevelagreementswithresearchprogramsgeneratingandcompilingthedataarenecessary for thefuture, and this aspect has been taken into account in designing the Copernicus AtmosphereMonitoringService.

2.4.1Prepareoperationalaccesstoground-basedremote-sensingproducts

ECMWFcontinuedtoacquireAERONETLevel1.5datainnear-real-time(onefileperdaycontainingall observations collectedbyNASA for thepreviousday) fromNASAGoddard. Theseobservationsarecurrentlyusedforverificationoftheglobalaerosolforecast.

2.4.2 Acquisition and pre-processing of ground-based remote-sensing observations in NRTanddelayedmode

Inaddition,theglobalvalidationactivitiesacquireregularlycolumndataavailableforaerosolswithintheframeworkoftheWMOGlobalAtmosphericWatchWorldDataCentreforaerosol(GAW-WDCA)

Page 26: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page26of148

hosted at NILU. The data acquisition is organised through NILUs web-interface EBAS(http://ebas.nilu.no).

AnautomatedscriptsystemdeliversNRT1-hourlydatawithintheframeworkofGAWsWorldDataCentre on Aerosols (GAW-WDCA-NRT) on the 15th of each month to MACC global validationactivities. The data delivered covers the beginning of the currentmonth and extends back to theprevious start of the year. It includes NRT Aerosol Optical Depth (AOD) data from 18 stationsworldwideusingasuntrackingsystemradiometer.ThestationsitesareindicatedinFigure2.2.Thedatacomponentsarelistedindetailinareport(D14.1).

Figure2.2.PositionofGAW_WDCA_NRTgroundbasedstationsdeliveringAODtoMACCglobalvalidationactivities.

2.5Researchsatelliteproducts

Theobjectiveofthisworkpackagetaskwastoidentifywaystooperationalizeandpre-processnewor improved retrieval satellite products and to evaluate the new satellite retrieval products usingground-baseddata. The ground-baseddata used inMACC-III for evaluationof the retrievalsweremainly from the ground-based EuropeanAerosol Research LidarNetwork (EARLINET). In addition,workunderOBS/MACC-III has proceed to showhowground-basedMulti-AxisDifferentialOpticalAbsorptionSpectroscopy(MAX-DOAS)instrumentscanbeusedasavaluabletoolforprovidingerrorcharacteristicsoffuturesatellite-basedproductsofatmosphericcomposition.

2.5.1Acquisitionofnewretrievalproductsfromresearchsatellitemissionsandnewproducts

ECMWFhasbeenresponsibletoarrangesuitablemethodsofacquisitionofresearchsatellitemissiondatawithdataproviderforinitialtestingandevaluationofthedatacapabilitiesforoperationaluse.OnecanmentioninparticularthenewPMAPaerosolproductdeliveredbyEUMETSAT.

2.5.2FurtherEvaluationofCALIPSOproductsagainstin-situNRTdata

Thisworkhas furtherevaluatedthecapabilitiesof spaceborneCALIOPdataonboard theCALIPSOsatellite. It has extended the inter-comparison work against the EARLINET ground based LIDARnetwork a publication in CALIPSO aerosol data and has added a new evaluation against theGeorgiaTechLIDARinAtlanta.TheresultshavebeenacceptedaspublicationintheMACC-II/-IIIACPspecialissueandarereportedinareport(D15).

Page 27: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page27of148

ThesuitabilityandqualityofCALIPSOLIDARproductswasevaluatedagainstEARLINETproducts inthecontextofassimilationintoforecastmodels inMACC-II/MACC-III. AmethodwasdevelopedtoconvertEARLINETbackscatterintotheattenuatedbackscattercoefficient,inordertointer-compareCALIOPandEARLINETcoincidentalmeasurements.Overthreeyears(fromNov.2010toDec.2012),48 CALIOP overpasses occurred within a 100km ground track offset distance from an operatingEARLINET station, resulting in 7405 data points used for the inter-comparison analysis. The inter-comparisonof the totalattenuatedbackscatterprofiles fromnear-real-timeCALIOPLevel1.5dataand convertedEARLINETdata showed fairly goodagreement.Onaverage, theCALIOPattenuatedbackscattervalueswereslightlyhigher (by3%) thantheEARLINETvalues.The levelofagreementbetweentheCALIOPandEARLINETattenuatedbackscattervalueswasinfluencedbythepresenceofaerosollayersinthePBLandFTandbytheaerosollayerheight.Atypeofdatafilteringwasusedtomitigate themultiple layers influence,and the filtering improvedtheagreementbetweenthe twodata sets in thePBL. Inaddition, splitting theaerosol layerheights into twocategoriesallowed todistinguish thedifferencesbetween thePBLand theFT (seeFigure2.3).Theanalysisalso showedthattheaerosol typesdetectedbyCALIOPwereconsistentwiththesourceoftheaerosolandthetransportmechanism.Aerosolsfromlocalsourcesweremainlydetectedintheboundarylayer,whilelong-rangetransportpollutionwasobservedintheFT.

Figure2.3.Comparisonof.CALIOPvsEARLINETtotalattenuatedbackscattercoefficientforCALIOPoverpassesoverEARLINETstationswithin100kmdistance.LeftfigurecolourcodeindicatesthegroundtrackdistancefromtheEARLINETstation.Rightfigurecolourcodeshowstheaerosollayeraltitude(From:Grigasetal.2015)

2.5.3 Validation of Earth Observation data and in particular starting to prepare for thevalidationoffutureCopernicusEOdata

Theworkunder this taskhasprepareda validationmethodology forCopernicusdata, particularlySentinel-5pandSentinel-3.Theobjectivehasbeentobuild-upavalidationmethodologyforsatellite-based trace gas retrievals using ground-based Multi-Axis Differential Optical AbsorptionSpectroscopy (MAX-DOAS). This is shownbasedon theexampleofnitrogendioxide (NO2)derivedfromdataacquiredby theOzoneMonitoring Instrument (OMI),and theGlobalOzoneMonitoringExperiment-2 (GOME-2). The findings of this study are directly relevant for assessing the errorcharacteristicsoftheupcomingTROPOsphericMonitoringInstrument(TROPOMI)foreventualuseinCAMS.

Page 28: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page28of148

TheresultsindicatethatMAX-DOASobservationsareveryvaluableasareferencepointforjudgingtheaccuracyofthetracegasretrievalsfromsatelliteinstruments.Oneissuetotakeintoaccountisthesomewhatdifferentspatialrepresentativenessofthetwoobservingsystems.Thisisparticularlyimportant forMAX-DOASstations located inornearurbanareaswherestrongspatialgradients inpollution levels exist. Such spatial gradients tend to be adequately captured by the MAX-DOASinstrumentsbutnotbythesatelliteinstrumentsduetothesignificantlycoarserspatialresolutionofthelatter.Assuch,strongbiasesbetweenthetwomeasurementssystemscanbeobservedatverypollutedsites.SomeresultsofthecomparisonarepresentedinFigure2.4.

TheresultsfurthermoreshowthattherearecurrentlyseveralMAX-DOASsitesprovidingdataovertime periods of 5 years or more. A direct validation of trends computed from satellite productsagainst trends computed at MAX-DOAS station is therefore starting to become reasonable. FirstcomparisonsatseveralsitesinEuropeandAsiaindicatereasonableagreementbetweenthetrendsbutalso show that thecurrent lengthsof the timeseriesat theMAX-DOASstationsarenotquitesufficientyettoaccuratelydeterminetrends.Theresultingsignificantuncertaintiesassociatedwiththe MAX-DOAS-derived trends therefore hinder a direct validation at this point. However, theseuncertaintiesareboundtoreducerapidlywithincreasingtimeserieslengthanditisanticipatedthatonly 1-2 additional years of data are required to obtain statistically significant results in pollutedregions. Once the MAX-DOAS time series are long enough for a significant number of stationsworldwide it is anticipated that they will provide the first method for directly validating thetropospheric NO2 trends obtained from satellite-based platforms without the need for indirectvalidation using models or similar techniques which introduce significant amounts of additionaluncertainty.

Figure2.4.ScatterplotsofmonthlyaveragesofthefullavailabledatasetsfromMAX-DOASandOMIat6MAX-DOASsites.Notethatbothaxesusealogarithmicscale

ThefindingsofthisstudyaredirectlyrelevantforassessingtheerrorcharacteristicsoftheupcomingTROPOsphericMonitoring Instrument(TROPOMI) foreventualuse inCAMS.Furtherdetailscanbefoundinareportentitled“Validationofsatellite-derivedNO2productsandtrendsusingMAX-DOASdata-PreparationforfutureCopernicusEOdata”(D15.2).

Page 29: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page29of148

2.6Feedbacktodataproviders

Theobjectiveofthisworkpackagewastoprepareacoordinatedfeedbacktotheobservationdataprovidersontheirdata,drawingfromtheexperiencegainedunderthedatauseintheproductionofMACC-IIIGlobalandRegionaldataassimilationsystems.

2.6.1Elaborationoffeedbacktodataproviders

InMACC-IIIwork under this task has extendeddeliverable reportD16.4 ofMACC-II by identifyingadditionalworkperformedduring2014-2015 in theuseofobservational satellitedataas input tothedataassimilationsystempre-operational inMACC-II/MACC-III. Itfocusesonrecentexperiencesfrom MACC-II/MACC-III, chiefly those described in papers submitted to the special issue of theACP/AMT/EESD/GMDjournals,“GMESMonitoringatmosphericcompositionandclimate,researchinsupport of the Copernicus/GMES atmospheric service”. It also discusses the outlook for futuremissions,drawingfromarecentreviewpaperonobservingsystemsimulationexperiments(OSSEs)forairqualitysatellitemissions(Timmermansetal.,2014).

Page 30: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page30of148

Page 31: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page31of148

3.Emissions(EMI)

ThegoaloftheEMIsub-projectistoprovidethedistributionofanthropogenicandbiogenicsurfaceemissionsattheglobalandregionalscalestotheMACC-IIIsub-projects.AspartofMACC-III,EMIhasworkedontheevaluationofglobalemissions,asafirststepinthedevelopmentofanewcommunityhistorical emission inventory. European regional emissions have been extended to cover morerecentyearsandmorecompounds.AmmoniaemissionsrelatedtoagricultureinEuropeandglobalbiogenicemissionshavebeenfurtherdevelopedandanalyzed.

EMI has collaborated with the different MACC-III sub-projects in order to ensure that all theproductsneededforeachsub-projectareavailableandsatisfactory.ThedatasetsdevelopedaspartofEMIaredistributedthroughtheECCAD(EmissionsofatmosphericCompounds&CompilationofAncillary Data) emissions database that was developed during the past few years. The databaseincludes tools to analyze and evaluate the surface emissions, and to compare them with otheravailabledatasets.

EMIhasevaluated thepossibilityofdevelopinga serviceprovidingoptimizedemissionsof carbonmonoxide.

EMI continued tohave strong linkswith theFIREproject,whichdevelopsandprovideanalysesofbiomass burning emissions: the emissions developed by the FIRE project are made available toMACC-IIIusersthroughtheECCADdatabase.

AllalongMACC-III,EMIhasdevelopedstrong interactionswith internationalprojectssuchasGEIA(Global Emissions Initiative) and AQMEII (EU-North American AQ Model Evaluation InternationalInitiative).TheemissionsdatasetsdevelopedaspartofMACC-IIIarealldistributedaspartofECCAD,whichistheemissionsdatabaseofGEIA.

3.1Europeananthropogenicemissions

DuringMACC-III and its extension, the TNO-MACC anthropogenic emission inventory was furtherdeveloped andmaintained to guarantee air qualitymodellers inMACC-III, but also outside of theproject, access to an updated state of the art, high resolution and consistent European emissioninventoryfor8priorityairpollutants(NOx,SO2,NMVOC,NH3,PM10,PM2.5,CO,CH4). InMACC-IIIthetime-seriesof the inventorywassubstantiallyexpandedtocoverallyearsbetween2000-2011andthespatialdistributionofemissionswas improved,providingabetterrepresentationofmajorcitiesintheemissiongrids.Atwo-stepapproachisfollowed:firstthetotalemissionsarequantifiedpercountry,sector,yearandpollutantandthenbrokendowntoadetailedsourcesectorlevel.Next,to produce model-ready emission grids, emissions are distributed spatially over Europe at aresolutionof1/8x1/16degrees(~7x7km)bylinkingthedetailedsourcesectorstosectorspecificspatial distributionmaps for area sourceswhile point sources are located at their exact location.Figure 3.1 illustrates both the resolution and domain of the model-ready grids as well as theemissiontrendforSO2between2000and2011.

Aspartof the furtherdevelopment inMACC-IIII, anew trendanalysiswasdone foremissions forInternationalshipping,asitwasrecognizedthattheEMEPshippingemissionsdatapreviouslyusedin MACC-II may no longer follow a realistic trend as important trends in shipping such as slowsteamingappearedtobemissing.BasedonreviewofavailabledataandexpertknowledgeatTNOandMet Norwaywe derived a new shipping emissions trend, including impacts of the economic

Page 32: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page32of148

crisis,foruseinourupdatedTNO-MACC-IIIemissioninventory.AsanillustrationinFigure3.1itcanbe seen that shippingemissionsof SO2declinedover time in theBaltic SeaandNorthSeadue toimplementationoftheSulphurEmissionControlArea(SECA)butnotonotherSeas.Thisnewsetofshippingemissionsdata is likely tobeadoptedbyEMEPaswell,althoughclearlymoreresearch isneededforthisimportantsource.

Figure3.1.DifferenceinemissionsofSO2betweentheyear2000-2011(positivevalues,ingreen,indicateanemissionreduction).

Another important development inMACC-III is the inclusion of CO2 emissions, next to the set ofpreviouslylistedairpollutants.WithinMACC-IIIregionalmodellersstartedtoworkonregionalCO2modellingandthereforeneedamodel-readygridofCO2emissionsdata.Thedatasetincludesasplitbetween fossil fuel CO2 andCO2 frombiofuels. This split is relevant for climate change studies asbiofuel CO2 is short-cycle carbon, a renewable fuel. For biogenic CO2 the largest contributor isresidential combustion, but an increasingly important contributor is the energy sector. Europeanfossil CO2 emissions remained fairly stable over the 2000-2011 period. An example of thedistributionofCO2emissionsinEuropein2009isgiveninFigure3.2.

Figure3.2.FossilCO2emissionsfromallsectorsinEuropefortheyear2009.

Page 33: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page33of148

AdetaileddocumentationoftheTNO-MACC_IIregionalEuropeanemissioninventorywaspublishedin the open source journal ACP by Kuenen et al. (2014). The TNO-MACC emission datawere alsousedintheUS-EPA/EC-JRCcoordinatedAirQualitymodelevaluationandintercomparisoninitiative(AQMEII) and a paper on comparison of European and American emissions data was published(Pouliotetal.,Atm.Env.,2015).AdocumentationoftheCO2emissioninventorydataisforeseeninthenearfuture.

ThepotentialofusingIASIsatellitedata(Clarisseetal.,2009)toprovidemesoscale informationtoconstrainEuropeanemissionsofammonia inorderto improve itsspatio-temporaldistributionhasbeenassessed.WeparticularlyfocusedontheweekofMarch8to15in2014,whenastrongpeakofparticulate pollution occurred in Europe. The comparison between the simulated CHIMEREconcentrations and those observed by IASI during the peak of pollution ofMarch 8 to 15 showsoverestimation of CHIMERE concentrations (Figure 3.3), mainly in France and southern Europe(Spain,Romania) andanunderestimationof the concentrations in centralGermany. It also showshighgeographicalvariabilityintheIASIobservationthatismissinginthemodeloutputs.

Figure3.3.ComparisonbetweentheNH3totalcolumnsobservedbyIASI(left)andthosesimulatedbythechemistry-transportmodelCHIMERE(right)in1015molec.cm-2,from8to15March2014.

To improve these ammonia emissions using IASI satellite data, a simple approach was followedbasedon theassumption thatNH3 is a short-lived species.Wehaveassumed that thedifferencesbetween CHIMERE and IASI are only due to the emissions prescribed in themodel. A correctionfactor corresponding to the relative difference between CHIMERE and IASI is applied at the gridpoint (0.5° x0.5°) inventory TNO/MACC to redistribute emissions. This correction leads to asignificantincreaseofemissionsinGermanyandareductioninFranceduringthisepisode.

3.2Globalanthropogenicemissions

AspartofthedevelopmentofanewCommunityHistoricalEmissionsInventory,EMIhasworkedonasystematiccomparisonofallpubliclyavailableglobalandregionalemissionsinventories.Thisworkfocuseson the1960-2015period: a largenumberof inventorieswere collected, and comparisonshavebeenmadefortheemissionsofCO,NOx,VOCs,SO2,BC,OC,PM10andPM2.5.Thegoalofthecomparisonistoidentifywhichdatasetsprovidethemostaccurateemissionsfortheperiodunderconsiderationforeachregion,andusetheseinventoriesasabasisforthenewcommunityhistoricalinventory. This new historical inventory will represent a new dataset developed by the scientificcommunitytobeusedinallglobalsimulationsperformedaspartofMACC/CAMS.

Page 34: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page34of148

Different types of inventorieswere used in this evaluation: global, regional aswell as inventoriesprovidingemissionsdataforasingleorseveralcountries.Bothgriddedandnon-griddedemissionsdatasetsareincludedinthiswork.

TheevaluationisusingtheMACCitydatasetasareference:theratioofallemissionstotheMACCityemissionsforeachregionandyearhasbeencalculated.AnexampleoftheresultsisshowninFigure3.4forCOinWesternandCentralEurope,aswellasChina.TheTNO-MACCinventoriesareshownaswellonthesefigures.ArathergoodagreementisfoundforallinventoriesinEurope,exceptfortheinventoriesdevelopedseveralyearsago (RETROandHYDE),whichshow loweremissions than theotherdatasetsinthe1960sands1970s.ComparedtoTNO-MACC-III,thetrendinMACCityemissionsis lower in bothWestern andCentral Europe. The trend in emissions from the ECLIPSE dataset ismuch larger than trends provided by the other inventories inWestern Europe, where ECLIPSE itfollowsthevaluesgivenbyRCPs2.6and4.5.

Much larger differences are found in the emissions in China:most of the recent datasets, exceptEDGAR 4.2, provide higher emissions. The MEIC emissions however, developed by local Chinesegroups having access to official emissions data, are very close to MACCity: however, the MEICemissions indicate a decrease in emissions since 2008, contrarily toMACCity which shows stableemissions.

Figure3.4.COemissionsinWesternEurope(topleft)andCentralEurope(middleleft)andratioswithMACCityforWesternEurope(topright),CentralEurope(middleright)andChina(bottomright).

Page 35: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page35of148

DuringMACC-III,theEMIsub-projectcooperatedwiththeTaskForceHemisphericTransportofAirPollution (HTAP) under the Convention on Long-Range Transboundary Air Pollution (CLRTAP) andcontributed to the harmonization and improvement of regional emission inventories as used in aglobalemissioninventorytosupportHTAPglobalandregionalscalemodellingefforts.Thisso-calledHTAP_v2.2emissioninventorycombinesthelatestavailableregionalinformationwithinacompleteglobaldatasetandisdocumentedindetailbyMaenhoutetal.(ACP,2015).

3.4.Naturalemissionsofhydrocarbons

As part of MACC-III, we have continued to evaluate natural emissions from the biosphere, andquantified the impactofusingdifferentemission inventorieson theatmospheric concentrationofchemical species. In collaboration with theMACC-III partner Juelich Research Center, a series ofsimulations using the MOZARTv3.5 (Model for OZone And Related Tracers) chemistry-transportmodel was used in order to evaluate how different levels of isoprene emissions impact theatmospheric composition. TheMOZARTmodel was forced by ECMWFmeteorology for the years2007-2008withanthropogenicemissionsfromtheMACCityinventoryandGFASv0biomassburningemissionsdevelopedaspartofMACC.Wetestedfiveemissioninventoriesofisoprene,includingtheMEGAN-MACCinventory,whichwascreatedduringtheMACC-IIproject.AcomparisonofmodeledandobservedtroposphericcolumnsofformaldehydeintwoselectedregionsisshowninFigure3.5.

Figure3.5.MonthlymeantroposphericcolumnsofformaldehydecalculatedbytheMOZARTmodelusingdifferentisopreneemissionsandobservedbytheGOME-2instrument,inEurope(left)andAustralia(right).

Resultsofthesimulationswerecomparedwithsatelliteobservationsofformaldehyde(CH2O),whichrepresentsoneofthemainproductsofisopreneatmosphericoxidation.Ingeneral,higherisopreneemissions lead to higher values of CH2O column. However, the relationship between isopreneemissionsandCH2Oconcentration isnot linear. There is a clearunderestimationof formaldehydetrop.columninEuropefromJanuarytoMarch.IsopreneemissionsinEuropeduringthisperiodarealmost equal to zero and the BVOC emissions are dominated mainly by monoterpene speciesemittedby coniferous trees. TheCH2Ounderestimation canbe an indicationof too low valuesofmonoterpeneemissionsinthemodelsimulations,butthisneedstobeanalyzedfurther.

Page 36: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page36of148

3.4Pilotstudy:aninversemodellingsystemofcarbonmonoxide

AspartofMACC-III, thenegative trendsof carbonmonoxide (CO) concentrationsobserved in therecentdecadebybothsurfacemeasurementsandsatelliteretrievalsovermanyregionsoftheglobewerestudied.Thesetrendsarenotwellexplainedbycurrentemissioninventories.Thecapabilityofatmospheric inversions to attribute the observed CO concentration decline was evaluated. Asophisticatedinversionsystemwasused,whichsimultaneouslyoptimizesthetwomainCOsources(surface emissions and atmospheric hydrocarbon oxidations) and the main CO sink (atmospherichydroxylradicalOHoxidation)byassimilatingobservationsofCOandofchemicallyrelatedtracers.Satellite CO column retrievals from Measurements of Pollution in the Troposphere (MOPITT),version6,andsurfacein-situmeasurementsofmethaneandmethyl-chloroformmolefractionswereassimilatedjointlyfortheperiodof2002-2011.Comparedtothepriorsimulation,theoptimizedCOconcentrationsshowedbetteragreementwithindependentsurfacein-situmeasurementsintermsofbothdistributionsand trends, as shown in Figure3.6 (Yinet al., 2015).At theglobal scale, theatmospheric inversion primarily interprets the CO concentration decline as a decrease in the COemissions,andfindsnoticeabletrendsneitherinthechemicaloxidationsourcesofCO,norintheOHconcentrations that regulate CO sinks. The latitudinal comparison of the model state withindependent formaldehyde (CH2O) columns retrieved from the Ozone Measurement Instrument(OMI) confirms the absenceof large-scale trends in the atmospheric source of CO. The global COemissions decrease by 17% during the decade,more than twice the negative trend estimated byemission inventories. The spatial distribution of the inferred decrease of CO emissions indicatescontributionsfrombothadecreaseinfossil-andbio-fuelemissionsoverEurope,theUSAandAsia,andfromadecreaseinbiomassburningemissionsinSouthAmerica,Indonesia,AustraliaandBorealregions.

Figure3.6.TimeseriesandspatialdistributionsofCOtotalcolumn(XCO).(a)TimeseriesofglobalmonthlymeanXCO.Theblack line represents satellite observation of MOPITTv6 XCO, the green (red) lines represent the prior (posterior)simulations.Solidlinesanddotedlinesrepresenttestswithdifferentconditions.(b)Distributionofmulti-yearmeanannualXCO of MOPITTv6 retrieval. (c) Mean annual difference between the prior simulation andMOPITT. (d) Mean annualdifferencebetweentheposteriorsimulationandMOPITT.

Page 37: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page37of148

3.5AccesstotheEMIdatasetsthroughtheECCADdatabase

All the datasets developed as part of EMI are publicly available through the ECCAD (Emissions ofatmospheric Compounds & Compilation of Ancillary Data) database, such as the most recentupdates of the MACCity global emissions, the TNO-MACC regional emissions, and the GFAS fireemissions.TheMEGAN-MACCnaturalemissionshavealsobeenincludedinECCAD.

The ECCADdatabase,whichdelivers theMACC-II emissionsdatasets hasnow reached2050usersfrom771worldinstitutions.ThisindicatesthattheemissionsproductsdevelopedaspartofMACC-IIarenowwidelyused,inEuropeandintherestoftheworld.References

3.6References

3.6.1Referencesforthissection

Clarisse, L., Clerbaux,C.,Dentener, F.,Hurtmans,D., andCoheur, P.:Global ammoniadistributionderivedfrominfraredsatelliteobservations,Nat.Geosci.,2,479–483,2009.

Granier,C.,Bessagnet,B.,Bond,T.,D’Angiola,A.,VanDerGon,H.D.,Frost,G.J.,Heil,A.,Kaiser,J.W.,Kinne,S.,Klimont,Z.,Kloster,S.,Lamarque,J-F.,Liousse,C.,Masui,T.,Meleux,F.,Mieville,A.,Ohara,T.,Raut, J-C.,Riahi,K., Schultz,M.G., Smith,S.J., Thompson,A.,VanAardenne, J.,VanderWerf,G.R.,VanVuuren,D.P.:Evolutionofanthropogenicandbiomassburningemissionsofairpollutantsatglobalandregionalscalesduringthe1980–2010period,ClimaticChange,109,162-190,2011.

Janssens-Maenhout,G.,Crippa,M.,Guizzardi,D.,Dentener,F.,Muntean,M.,Pouliot,G.,Keating,T.,Zhang,Q.,Kurokawa,J.,Wankmüller,R.,DeniervanderGon,H.,Klimont,Z.,Frost,G.,Darras,S.,andKoffi,B.:HTAP_v2:amosaicofregionalandglobalemissiongridmapsfor2008and2010tostudyhemispherictransportofairpollution,Atmos.Chem.Phys.Discuss.,15,12867-12909,doi:10.5194/acpd-15-12867-2015,2015.

Kuenen,J.J.P.,A.J.H.Visschedijk,M.Jozwicka,andH.A.C.DeniervanderGon,

TNO-MACC_II emission inventory: a multi-year (2003–2009) consistent high-resolution Europeanemission inventory for air quality modelling, Atmos. Chem. Phys., 14, 10963-10976,doi:10.5194/acp-14-10963-2014,2014.

LamarqueJ.F.,BondT.,EyringV.,GranierC.,HeilA.,KlimontZ.,LeeD.,LiousseC.,MievilleA.,OwenB.,SchultzM.G.,ShindellD.,SmithS.J.,StehfestE.,vanArdenneJ.,CooperO.R.,KainumaM.,Mahowald N.,McConnell J.R., Naik V., Riahi K., and van Vuuren D.P., Historical (1850–2000)gridded anthropogenic and biomass burning emissions of reactive gases and aerosols:methodologyandapplication,ACP10,7017-7039,2010.

Maignan, F.,Bréon, F.-M.,Chevallier, F.,Viovy,N.,Ciais,P.,Garrec,C., Trules, J., andMancip,M.:EvaluationofaGlobalVegetationModelusingtimeseriesofsatellitevegetationindices,Geosci.ModelDev.,4,1103-1114,doi:10.5194/gmd-4-1103-2011,2011.

Pouliot G., H.A.C. Denier van der Gon, J. Kuenen, J. Zhang,M.Moran, P.Makar, Analysis of theEmission Inventories and Model-Ready Emission Datasets of Europe and North America forPhase2oftheAQMEIIProject,Atmos.Env.,doi:10.1016/j.atmosenv.2014.10.061,2014.

Page 38: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page38of148

Yin, Y., F. Chevallier, P. Ciais, G. Broquet, A. Fortems-Cheiney, I. Pison, andM. Saunois: Decadaltrends in global CO emissions as seen by MOPITT. Atmos. Chem. Phys. Discuss., 15, 14505-14547,doi:10.5194/acpd-15-14505-2015,2015.

3.6.2PublicationswithinMACC-III

Im,U.,Bianconi,R.,Solazzo,E.,Kioutsioukis,I.,Badia,A.,Balzarini,A.,Baro,R.,Bellasio,R.,Brunner,D., Chemel, C., Curci, G., Denier van der Gon, H.A.C., Flemming, J., Forkel, R., Giordano, L.,Jimenez-Guerrero, P., Hirtl, M., Hodzic, A., Honzak, L., Jorba, O., Knote, C., Makar, P.A.,Manders-Groot, A., Neal, L., Perez, J.L., Pirovano, G., Pouliot, G., San Jose, R., Savage, N.,Schroder,W.,Sokhi,R.S.,Syrakov,D.,Torian,A.,Tuccella,P.,Werhahn,K.,Wolke,R.,Yahya,K.,Zabkar, R., Zhang, Y., Zhang, J., Hogrefe, C., Galmarini, S., Evaluation of operational online-coupled regionalairqualitymodelsoverEuropeandNorthAmerica in thecontextofAQMEIIphase 2. Part II: Particulate matter. Original Research Article, Atmospheric Environment,doi:10.1016/j.atmosenv.2014.08.072,2014.

Im,UlasRobertoBianconi,EfisioSolazzo,IoannisKioutsioukis,AlbaBadia,AlessandraBalzarini,RocíoBaró,RobertoBellasio,DominikBrunner,CharlesChemel,GabrieleCurci, JohannesFlemming,RenateForkel,LeaGiordano,PedroJiménez-Guerrero,MarcusHirtl,AlmaHodzic,LukaHonzak,Oriol Jorba, Christoph Knote, Jeroen J.P. Kuenen, Paul A.Makar, AstridManders-Groot, LucyNeal, Juan L. Pérez, Guido Pirovano, George Pouliot, Roberto San Jose, Nicholas Savage,WolframSchroder,RanjeetS.Sokhi,DimiterSyrakov,AlfreidaTorian,PaoloTuccella,JohannesWerhahn,RalfWolke,KhairunnisaYahya,RahelaZabkar,YangZhang, JunhuaZhang,ChristianHogrefe, Stefano Galmarini, Evaluation of operational on-line-coupled regional air qualitymodels over Europe and North America in the context of AQMEII phase 2. Part I: Ozone,AtmosphericEnvironment,doi:10.1016/j.atmosenv.2014.09.042,2014

Fuzzi,S.,U.Baltensperger,K.Carslaw,S.Decesari,H.DeniervanderGon,M.C.Facchini,D.Fowler,I.Koren,B.Langford,U.Lohmann,E.Nemitz,S.Pandis,I.Riipinen,Y.Rudich,M.Schaap,J.Slowik,D.V.Spracklen,E.Vignati,M.Wild,M.Williams,andS.Gilardoni,Particulatematter,airqualityandclimate:lessonslearnedandfutureneeds,Atmos.Chem.Phys.Discuss.,15,521-744,2015

Janssens-Maenhout,G.,Crippa,M.,Guizzardi,D.,Dentener,F.,Muntean,M.,Pouliot,G.,Keating,T.,Zhang,Q.,Kurokawa,J.,Wankmüller,R.,DeniervanderGon,H.,Klimont,Z.,Frost,G.,Darras,S.,andKoffi,B.:HTAP_v2:amosaicofregionalandglobalemissiongridmapsfor2008and2010tostudyhemispherictransportofairpollution,Atmos.Chem.Phys.Discuss.,15,12867-12909,doi:10.5194/acpd-15-12867-2015,2015.

Kuenen,J.J.P.,A.J.H.Visschedijk,M.Jozwicka,andH.A.C.DeniervanderGon,

TNO-MACC_II emission inventory: a multi-year (2003–2009) consistent high-resolution Europeanemission inventory for air quality modelling, Atmos. Chem. Phys., 14, 10963-10976,doi:10.5194/acp-14-10963-2014,2014.

Pouliot G., H.A.C. Denier van der Gon, J. Kuenen, J. Zhang,M.Moran, P. Makar, Analysis of theEmission Inventories and Model-Ready Emission Datasets of Europe and North America forPhase2oftheAQMEIIProject,Atmos.Env.,doi:10.1016/j.atmosenv.2014.10.061,2014.

Sindelarova,K.,Granier,C.,Bouarar,I.,Guenther,A.,Tilmes,S.,Stavrakou,T.,Müller,J.-F.,Kuhn,U.,Stefani, P., andKnorr,W.:Global dataset of biogenicVOCemissions calculatedbytheMEGAN

Page 39: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page39of148

modeloverthelast30years,Atmos.Chem.Phys.Discuss.,14,10725–10788,doi:10.5194/acpd-14-10725-2014,2014.

Vilardell, G., M. T. Pay, F. Martínez, A. Soret, H. Denier van der Gon, J.M. Baldasano, Inter-comparison between HERMESv2.0 and TNO-MACC-II emission data using the CALIOPE airquality forecastsystem(Spain),AtmosphericEnvironment,Volume98,December2014,Pages134-145,2014.

Yin, Y., F. Chevallier, P. Ciais, G. Broquet, A. Fortems-Cheiney, I. Pison, and M. Saunois: Decadaltrends in global CO emissions as seen by MOPITT. Atmos. Chem. Phys. Discuss., 15, 14505-14547,doi:10.5194/acpd-15-14505-2015,2015.

3.6.3Presentationsatinternationalmeetings

Darras,C.,C.Granier,C,Liousse,D.Boulanger,E.Enriquez,K.Sincelarova,T.Doumbia,TheECCADDatabase: Access to Global and Regional Emissions Data, 2015 EPA International Conference"AirQualityChallenges:tacklingtheChangingFaceofEmissions",SanDiego,USA,April2015.

Darras,S.,C.Granier,C.Liousse,E.Enriquez,D.BoulangerandAudeMieville,Access toEmissionsDistributionsandRelatedAncillaryDatathroughtheECCADdatabase,EGUGeneralAssembly,Vienna,Austria,April2015.

Frost,G., L. Tarrason,C.Granier, P.Middleton, TheGlobal Emissions Initiative (GEIA):working forbetteremissionsinformationacrosstheworld,2015EPAInternationalConference"AirQualityChallenges:tacklingtheChangingFaceofEmissions",SanDiego,USA,April2015.

Frost, G.J., C. Granier, S. Darras, T. Doumbia, S.W. Kim, B. Hassler, K. Sindelarova, T. Ryerson,M.Trainer,Improvingknowledgeofsurfaceemissionsusingobservationsfromsatelliteandaircraftcampaigns,ESAAtmos2015conference"Advances inAtmosphericScienceandApplications",Heraklion,Greece,June2015.

Granier, C., Sindelarova, K., T. Doumbia, S. Darras, H. Denier van der Gon, G. Frost, Changes inanthropogenic surface emissions during the past decades: comparisons between differentglobal and regional inventories, 2015 EPA International Conference "Air Quality Challenges:tacklingtheChangingFaceofEmissions",SanDiego,USA,April2015.

Granier,C., I.Bouarar,A.Colette, T.Doumbia, L. Emmons,A.Hilboll,A.Richter,K. Sindelarova, S.Tilmes, H.Worden, Evolution of the chemical composition of the atmosphere over the pastdecades:comparisonsbetweenchemistry-climatemodelsimulationsandsatelliteobservaitons,ESA Atmos 2015 conference "Advances in Atmospheric Science and Applications", Heraklion,Greece,June2015.

Janssens-Maenhout,G.,M.Crippa,D.Guizzardi,E.Schaaf,M.Muntean,F.Dentener,K.Sindelarova,C. Granier, Scenarios over the past 3 decades: air quality impact of European legislation,AmericanGeophysicalUnionFallmeeting,SanFrancisco,USA,December2014.

Kuenen,J.,A.Visschedijk,H.DeniervanderGon,EuropeanEmissionInventorywithFuelsplitdetail,Fairmodetechnicalmeeting,http://fairmode.jrc.ec.europa.eu/Aveiro,Portugal,June2015.

Liousse,C.,E.M.Assamoi,E.T.N'Datchoh,T.Doumbia,L.Granier,P.Criqui,L.Roblou,C.Granier,A.Konare, R. Rosset, G. van der Werf, and J. Kaiser, African combustion emission inventory:specificitiesanduncertainties,SeventhInternationalSymposiumonnon-CO2GreenhouseGases(NCGG7),Amsterdam,TheNetherlands,November2014.

Page 40: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page40of148

Sindelarova, K., C. Granier, T. Stavrakou, J.-F. Muller, O. Stein, M. Schultz, A. Guenther, Spatio-temporalvariabilityofbiogenic isopreneemissionsand their impactonatmosphericchemicalcomposition,IGACconference,Natal,Brazil,September2014.

Page 41: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page41of148

4.FireDataAssimilation(FIR)

The FIR subproject provides global estimates of smoke constituent fluxes from open biomassburning,e.g.forestfires,savannahfires,agriculturalwasteburning.TheseestimatesareprovidedinnearrealtimetoallatmosphericmodellingsystemsofMACC-IIIandagrowinguserbaseoutsidetheproject. They constitute boundary conditions for atmospheric analyses and forecasts thatcomplementtheinformationfromsatellitedataassimilationandareessentialfortheaccuracynearthesurface(Huijnenetal.2012).

During MACC-III, FIR has ported its two real time GFAS production chains to the newsupercomputers of ECMWF, continued its validation activities in collaboration with the globalatmospheric systems of CAMS, and prepared further developments of GFAS. The FIR productionchainsnowcomprise:

• GFASv1.2dailyproductionofFRP,emissionandinjectionheightfields• early-deliveryversionofGFASv1.2,whichisavailableat20UTCofthesameday,forusein

Europeanairqualityforecasts• Fire product generation from the GOES-East and –West satellites in operational

infrastructureatIPMAusingadedicatedalgorithmfocusedonFRPanddevelopedbyKCL.

The Linux PC version of GFAS, which is currently designed to operate within the ECMWFinfrastructure,isbeingadaptedtotheoperationalinfrastructureatIPMAandtoastand-aloneLinuxPCoutsideECMWF.Thisdevelopmentwillallowrapidtransitionofresearchupdatestooperationsinthe future. Strict version control is enforced with a GIT software repository. Independence ofCommercial Off-The-Shelf (COTS) software tools is achieved through the use of GIT, python andFortran.

Furthermore, FIR continued to spend considerable effort in supporting theGFASuser basewithinMACC-IIIandbeyond.FIRhasalsocontributedfurthertothefireproductinter-comparisoninESA’sFire-cci project. As in the years before, GFAS describes fire disturbance in NOAA’s State of theClimatereportfor2014(Kaiser&vanderWerf2015).

Theinteractionwiththeinternationalscientificcommunitywasstrengthenedthroughparticipationinseveral internationalscientificconferences, inGEIAand iLEAPSsteeringcommitteemeetings,aswellasthroughcollaboration inseveralstudiesofscientificusers.ThedisseminationofGFASfromGEIA’sECCAdatabaseison-going.

Akey risk to theFIR fireemissionservice is its strongdependenceon theMODIS instrumentson-boardtheTerraandAquasatellites.FIRhaspreparedforearlyadoptionofSentinel-3FRPproductsthroughitscontributiontotheSentinel-3LandCal/Valteam,andtheseproductshavebeen(underaseparatecontract)prototypedbyKCL.FIRhasalsoanalysedtheerrorcharacteristicsoftheexistingNPP-VIIRSfireobservationstoprepareusingtheupcomingfireproductsfromthissatellite.FIRhasfurthermore, analysed the representability of the diurnal fire cycle from LEO observation inpreparation of a version with 1-hour resolution, which could then also ingest GEO observationswhereavailable.

Concerning retrospective emissions, FIR has consistently reprocessed the period 2000-2014 withversion1.2ofGFAS,andcontributedtoversion4oftheGFEDinventory,whichwasreleaseon1July2015.

Page 42: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page42of148

4.1Fireemissionserviceprovision

4.1.1FRPservicechaininnearrealtime

ThroughoutMACC-III,IPMAhasgeneratedFRPproductsusingtheKCLalgorithmfromtheGOES-Eastand GOES-West observations in native satellite resolution in real time. The products are derivedfrom the satellite radiance products acquired from UCAR and the operational meteorologicalforecastsacquiredfromECMWF.TheyaredisseminatedbyIPMAwithin4hoursoftheobservationtime. The IPMAproduction conforms to the “Fire Radiative Power –GOES ProductUserManual”(Trigoetal.2014/MACC-IIDeliverableD_32.1).IPMAarchivestheGOESFRPproductsandallinputdata. ECMWF archives the GOES FRP products and lists them publicly in the MACC-II productcatalogue.TheGOESFRPserviceisthus“operational”exceptfor24/7humansupportandrealtimequality control of the product. The operational monitoring of the production, which includesverification of processed pixels and product timeliness, has continued throughout MACC-III anddemonstratesthestabilityoftherealtimeservice,cf.Figures4.1&4.2.

Figure4.1.NumberofFRPfilesgeneratedduringeachmonthoftheMACC-IIIprojectdurationfromGOES-West(toppanel)andGOES-East(bottompanel),within4hourtimelinessandafter.NorthernHemisphere(NH)coverageisavailableevery30minutes,asignificantlyhighertemporalfrequencythanSouthernHemisphere(SH)andFullDisk(FD)coverage,explainingthedifferenceinfilenumbers.(graphicsbyJ.Macedo,IMPA).

Page 43: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page43of148

Figure4.2.SnapshotofthesystemmonitoringtoolatIPMAshowingthenumberofprocessedpixelspersatelliteandarea.The operator can choose a range of display options/information. Red dots correspond tomissing input datawithin therequired timeliness. In the case of GOES-E (right panels), the red line at the beginning of June 2015 indicates a serviceinterruption,whichcouldbetracedtoa failure inthedistributionofrawGOES-E inputdata.Variations inthenumberofprocessedpixelscorrespondto(mostlyperiodic)changesintotalareacoverageofeachwindow(NH:NorthernHemisphere;FD:FullDisk;SH:SouthernHemisphere).(graphicsbyJ.Macedo,IPMA).

4.1.2Fireemissionservicechain,realtime

The daily real time fire emission service GFASv1.2 with 0.5º and 0.1º resolutions, has beenmaintainedandrunthroughoutMACC-III.Anoverviewofthecontinentalanddailyfiredistributionisgiven in Figure 4.3. In September 2014 an additional area over Icelandwasmasked following theoutbreakofavolcanothathadnotpreviouslybeenobservedasthermalanomaly.Furthermore,theGFASv1.2 production was ported to the new supercomputer at ECMWF in September 2014, andMARSarchivingiscopiedtotheclassdesignatedfortheCopernicusservicesinceMay2015.Figure4.4depictstheroutinemonitoringoftheGFASproductiontimelinessin2015.ItshowsthatthedatagenerationandprovisiontotheglobalatmosphericCAMSsystems(“model”)waslessthanoncepermonthbehind the target time,while thearchiving finishedafter the target timemore frequently.ThearchivingdelaysoccurringsinceMayarenowunderinvestigation.

Followinguserrequests,twoadditionalfeaturehavebeenproducedthroughoutMACC-III:

• Hourly gridded representations of the FRP observations from Terra MODIS, Aqua MODIS,MeteosatSEVIRI,GOES-EastImagerandGOES-WImager.

• AcopyofGFASv1.2withshiftedtimewindowhasbeenavailableat2000UTCforEuropeanairquality forecasting inMACC-III. This product is a suitable estimate for fire emission in Europeduring the same day because it takes the MODIS observations obtained during the earlyafternoonintoaccount.

Page 44: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page44of148

FIR has collaborated closely with AER, GHG and GRG to implement the new injection heightinformationintheatmosphericmodelandtestitsimpactontheatmosphericserviceaccuracy.

GFASdatahavebeendisseminatedthroughtheECWMFdisseminationftpserver,theMACC-IIIwebsite, the JOINserveratFZ Jülich, theECCADserver,MARSaccessbyusers,andpersonal contacts.Theftpserverprovidesthemostreliableandtimelydissemination.Itrequiredregistrationandhasbeenusedbyelevenusers:

Name/ID e-mail CountryUlrich.Roemer [email protected] agata.hoscilo [email protected] plcharles.m.ichoku [email protected] uschristopher.gan [email protected] sgjesus.sanmiguel [email protected] itluke.ellison [email protected] usmasayuki.takigawa [email protected] jpsunhee.lee [email protected] auxin.lin [email protected] fryiannis.proestos [email protected] cyyusheng.shi [email protected] jp

Furthermore,GFASdatahavebeenaccesseddirectlyfromtheMARSarchivebyanumberofusersfrom the ECMWFmember states and from the archive at FZ Jülich and GEIA’s ECCAD archive inFrance.Ithasbeendownloaded63timesfromECCADby:

1. Armando|Rodriguez|[email protected]. Ashfold|Matthew|[email protected]. Auby|Antoine|[email protected]. Baldassarre|Giuseppe|[email protected]. Basu|Sourish|[email protected]. Bauwens|Maite|[email protected]. Benjamin|Poulter|[email protected]. Berezin|Engeny|[email protected]. Bhardwaj|Piyush|[email protected]. Breil|Romaric|[email protected]. Chandra|Prafulla|[email protected]. Cottle|Paul|[email protected]. Darras|Sabine|[email protected]. Dawidowski|Laura|[email protected]. Desimone|Francesco|[email protected]. Ellison|Luke|[email protected]. Emmons|Louisa|[email protected]. F|F|[email protected]. Frost|Gregory|[email protected]. Granier|Claire|[email protected]. Hayabuchi|Yuriko|[email protected]. Huneeus|Nicolas|[email protected]. Hyer|Edward|[email protected]. Ikeda|Kohei|[email protected]. Imran|Girach|[email protected]. Jo|Duseong|[email protected]. Kaiser|Johannes|[email protected]

Page 45: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page45of148

28. Kaul|Daya|[email protected]. Kurata|Gakuji|[email protected]. Lasko|Kristofer|[email protected]. Leon|Jean-francois|[email protected]. Lin|Xin|[email protected]. Lorentz|K|[email protected]. Maksyutov|Shamil|[email protected]. Mao|Huiqin|[email protected]. Mathibela|Kabelo|[email protected]. Naoya|Kuramoto|[email protected]. Oliva|Patricia|[email protected]. Paunu|Ville-veikko|[email protected]. Pereira|Gabriel|[email protected]. Praznikar|Jure|[email protected]. Raffuse|Sean|[email protected]. Reddington|Carly|[email protected]. Roblou|Laurent|[email protected]. Satoru|Koizumi|[email protected]. Schreier|Stefan|[email protected]. Sindelarova|Katerina|[email protected]. Soares|Joana|[email protected]. Souri|Amirhossein|[email protected]. Sugeng|Nugroho|[email protected]. Takigawa|Masayuki|[email protected]. Tanaka|Taichu|[email protected]. Thanonphat|Boonman|[email protected]. Tian|Chongguo|[email protected]. Turquety|Solene|[email protected]. Veira|Andreas|[email protected]. Wheida|Ali|[email protected]. Whitburn|Simon|[email protected]. Yang|Zhifeng|[email protected]. Yim|Steve|[email protected]. Young|Jenny|[email protected]. Yun|Kim|[email protected]. Zhang|Feng|[email protected]

The MACC-III has dealt with an increasing number of enquiries on GFAS. Additionally, personaladvicebytheFIRteamanddatahavebeenprovidedtothefollowing:

1. AshuDastoor,EnvironmentCanada,Canada2. ChristosGiannakopoulos,TheCyprusInstitute,Cyprus3. KevinDelaney,EnvironmentIreland,Ireland4. ShamilMaksyutov,NIES,Japan

Page 46: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page46of148

Figure4.3.Continental-scalefireactivityfor1March2000to8July2015.Anextremeonsetofthe2015fireseasoninborealNorthAmericaisevidentfromthetopleftpanel.GFASv1.2datapublishedathttp://atmosphere.copernicus.eu/fire.

Page 47: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page47of148

Figure4.4.RealtimeavailabilityofMACC-IIIGFASv1.2during2015.TimeinUTC.

4.1.3Retrospectivefireemissions

Theyears2003to2014werereprocessedwithGFASv1.2togenerateconsistenttimeseriesoffiredistributions,emissionsand injectionheights for climateapplications.The timeserieswasalreadyusedinNOAA’sBAMSStateoftheClimatein2014report(Kaiser&vanderWerf2015).Figure4.5reproduces the2014 fire anomalymap from this report. Thedatahas also alreadybeenused forstudyingtheeffectofinjectionheightsinclimatemodelsbyVeiraetal.2015.

Furthermore, Version 4 of the GFED inventory was finished with MACC-III support and madepublicallyavailableon1July2015.

Figure4.5.2014globalfireanomalyfromNOAA’sBAMSStateoftheClimatein2014report,basedonGFASv1.2data.

Page 48: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page48of148

4.1.4Validation

The FIR team cooperates intensivelywithinMACC-II andwith project-external users to assess theaccuracyof the fireemissionproducts.TheGFASemissionsarevalidated incollaborationwiththeGFASuserseitherby comparingdirectly toother fireemission inventoriesorby comparingGFAS-basedplumesimulations toatmosphericobservations.The regular validation reportsproducedbytheVAL SP thus regularly contain case studiesof smokeplumes. Furthermore,GFAS is commonlyused and compared in the scientific literature. FIR has in particular tested the consistency of the0.5degand0.1degproductsofthereal-timeGFASproduction,aswellastheconsistencyofthereal-timeandretrospectiveproductionsofGFASv1.2(D31.1andD31.2).

FIR has also cooperatedwith theAER,GRGandGHG sup-projects in analyses of specific fire casestudies.ThevalidationofthenewinjectionheightinformationinGFASv1.2isofparticularinterest.Figure 4.6 exemplarily reproduces a comparison of LIDAR observations with MACC aerosolsimulationswithandwithoutGFASinjectionheightsfromD32.1.

Figure 4.6. Comparison of SAEC4RS Lidar observations of aerosol extinction (top)withMACCmodel simulationswithout(middle)andwith(bottom)fireemissioninjectionheightsfromGFAS.(copiedfromD32.1).

Page 49: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page49of148

4.2Fireemissionserviceenhancement

Thisworkpackageaimsatmajornewscientificdevelopmentsandscientificupgradesoftheexistingproducts. These developments provide the basis for major upgrades to the service provided inWP31.TheinvestigationsperformedduringMACC-IIIaredocumentedindetailinareport(D32.1).

4.2.1InvestigationofconsistencybetweenFRP-basedandGFEDsystem

Conversion factors should be re-calculated with the new version 4 of GFED. However, ourinvestigationsshowthattheseconversionfactorsw.r.t.GFEDexhibitunexplainedvariability,whichmightpartiallybeduetoinaccuraciesinGFED.Therefore,investigationswiththeaimofconstrainingthe conversion factors with the atmospheric observations assimilated in the global CAMS systemhavestartedincollaborationwithGHG.

4.2.2Useofdynamicemissionfactors

Therewere no user requests for additional species and no new insights into the use of dynamicemissionfactorscouldbegained.

4.2.3Injectionheightparameterisation

ThePlumeRiseModel(PRM)thatisusedintheGFASproductionchainhasbeenfurtheroptimisedanddocumented.Variousvalidationeffortshavetakenplace,e.g.thecomparisonshowninFigure4.6.

4.2.4FRPassimilation

It is recommended to implement a version of GFASwith 1-hour time step size in preparation ofingesting the high-temporal-resolution observations from geostationary platforms. This requires aparameterisedmodelofthediurnalcycle,whichcanbeusedbasedonMODISobservationsalone.We find that a parameterisation with a Gaussian function on top of a constant background issuitablewhenthepeakhourandpeakwidtharedeterminedapriorifromaclimatology.

InordertopreparefortheassimilationofVIIRSfireobservations,VIIRSdatahavebeenacquiredandKCLhavedevelopedatestalgorithmtodeliverFRPinformationfromtheVIIRSI-band(highspatialresolution)observations.ThealgorithmisbasedontheofficialversionfromNOAA/NASA/UniversityofMaryland,withsomeadditionstofocusperformanceanalysisonagriculturalfireareaswherefiresare highly numerous but typically small in size and where false alarms are also known to be aconcern. The resulting data are found to be substantially better at detecting small fires than isMODIS.It isthusexpectedthatthecontributionfrom,e.g.,agriculturalwasteburningwill increasesignificantlyinGFASonceVIIRSFRPproductsbecomeavailableinrealtimeandareused.

4.2.5Fireactivityforecasts

Forecasting the fire activity based on information from numerical weather forecasts (viameteorologically-driven fire danger indices) remains challenging, in part because there are manyfactors besidemeteorology that influence fire behaviour (e.g. fuel availability, landscape pattern,human ignition and suppression efforts…), and the significance of these varies widely between

Page 50: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page50of148

biomesandlocations.However,ourinvestigationspointtosomeforecastingskillfortheevolutionofexceptionally largefireswhencomponentsfromtheCanadianFireWeatherIndexsystemareused(boththefinal'FWI'metric,butoftensomeoftheunderlyingcomponentssuchastheInitialSpreadIndex,ISI)-seeanexampleinFigure4.7.

Figure4.7.ExampleoftherelationshipbetweenmeteorologicalfiredangerindicesandfireactivityfortheextremeRussianpeatandvegetationfiresofAugust2010thatburnedoutsideMoscow.ThefirelocationshowndemonstratesanFRPpeak(reddashed line)thatcoincideswiththemainpeak inthe ISI (left)andFWI(right)metricsoftheCanadianFireWeatherIndex(greylines),andsometemporalpatternsintheFRPdataalsoappeartoreflectthoseinfireactivity,albeitwithsomeinconsistentlag.

TheFWIcomponentsare readilyavailableatECMWFfromtheEuropeanFloodAwarenessSystem(EFAS) and could be deployed within GFAS to adjust the assumed persistence of fires in bothforecasts andwhere observations are unavailable currently (e.g. gaps in data). In themean time,persistencefor5daysremainstherecommendationfortheglobalfiredistributionandpersistencefor2-3daysisrecommendedfortheEuropeanfireemissions.

4.3Fireemissionservicecoordination

Thework inFIR iscloselycoordinatedwiththerelateddevelopmentswithEMIandtheothersub-projectofMACC-III.

4.3.1ManagementofFIRsub-project

Closecollaborationinthesubprojectismaintainedbyregularskypeteleconferences,frequentemailexchange and telephone calls as required. Individual team members meet and discuss FIRdevelopments at MACC-III General Assemblies, various scientific conferences and other externalmeetings.

Close collaboration with the other MACC-III SPs has been maintained through, amongst others,participationintheManagementBoardandtheGeneralAssembly.

4.3.2Interfacetowiderscientific,operationalandusercommunity

External relationships have been strengthened through presentations at, amongst others, theESA/EUMETSATSentinel-3Cal/ValWorkshopinDecember2014,theAGUFallMeetinginDecember

Page 51: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page51of148

2014 and the EGU General Assembly in April 2015. Furthermore,MPG continues to co-chair theInterdisciplinary Biomass Burning Initiative (IBBI, http://www.dl.mpic.de/ibbi) of WMO, IGAC andiLEAPS, and participate in the Steering Committee of IGAC’s Gobal Emissions Initiative (GEIA,http://geiacenter.org). Occasional contacts aremaintainedwith the JRC EFFIS and theGlobal FireMonitoring Centre (GMFC, http://www.fire.uni-freiburg.de) in order to ensure complimentarity ofthedevelopmentsandcapitaliseonupcomingopportunitiesforsynergies.

4.4References

BelowisthelistofpublicationsbyFIRpersonnelduringMACC-III.

Andela, N., Kaiser, J. W., van derWerf, G. R., andWooster, M. J. (2015). New fire diurnal cyclecharacterizations to improve fire radiative energy assessments made from low-Earth orbitsatellitessampling.AtmosphericChemistryandPhysicsDiscussions,15(6):9661–9707.

Archer-Nicholls, S., Lowe,D.,Darbyshire,E.,Morgan,W.T.,Bela,M.M.,Pereira,G., Trembath, J.,Kaiser, J. W., Longo, K. M., Freitas, S. R., Coe, H., and McFiggans, G. (2015). CharacterisingBrazilian biomass burning emissions using WRF-Chem with MOSAIC sectional aerosol.GeoscientificModelDevelopment,8(3):549–577.

Evangeliou,E.,Y.Balkanski,A.Cozic,W.M.Hao,F.Mouillot,K.Thonicke,R.Paugam,S.Zibtsev,T.A.Mousseau, R.Wang, B. Poulter, A. Petkov, C. Yue, P. Cadule, B. Koffi, J.W. Kaiser, and A. P.Møller2015.FireevolutionintheradioactiveforestsofUkraineandBelarus:futurerisksforthepopulation and the environment. Ecological Monographs 85:49–72.http://dx.doi.org/10.1890/14-1227.1

Gonzi,S., Palmer,P.I., Paugam,R., Wooster,M., and Deeter,M.N.: Quantifying pyroconvectiveinjection heights using observations of fire energy: sensitivity of spaceborne observations ofcarbonmonoxide,Atmos.Chem.Phys.,15,4339-4355,doi:10.5194/acp-15-4339-2015,2015.

Ichoku,C.,Ellison,L.,Yue,Y.,Wang,J.,andKaiser,J.W.(2015).Fireandsmokemeasurementandmodeling uncertainties. In Webley, P., editor, Natural Hazard Uncertainty Assessment, AGUBooks.AGU.accepted.

Inness,A.,Benedetti,A.,Flemming,J.,Parrington,M.,Kaiser,J.W.,andRemy,S.(2015).TheENSOsignal inatmosphericcompositionfields:Emissiondrivenversusdynamically inducedchanges.Atmos.Chem.Phys.Discuss.,(9):13705–13745.

Kaiser, J.W.,Benedetti,A.,Chevallier,F.,Flemming, J., Inness,A.,andPeuch,V.-H. (2015). [Globalclimate]climatemonitoringmeetsairqualityinCAMS[in”StateoftheClimatein2014”].BAMS.inpress.

Kaiser, J. W. and van der Werf, G. R. (2015). [Global climate] Biomass burning [in ”State of theClimatein2014”].BAMS.inpress.

Keywood, M. and Kaiser, J. (2014). IGAC/iLEAPS/WMO third workshop for the interdisciplinarybiomassburninginitiative.IGACNewsletter,52:8.

Konovalov, I.B.,Berezin,E.V.,Ciais,P.,Broquet,G.,Beekmann,M.,Hadji- Lazaro, J.,Clerbaux,C.,Andreae,M.O.,Kaiser,J.W.,andSchulze,E.-D.(2014).ConstrainingCO2emissionsfromopenbiomassburningbysatelliteobservationsofco-emittedspecies:amethodanditsapplicationtowildfiresinSiberia.Atmos.Chem.Phys.,14(19):10383–10410.

Page 52: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page52of148

Remy,S.andKaiser,J.W.(2014).Dailyglobalfireradiativepowerfieldsestima-tionfromoneortwoMODISinstruments.Atmos.Chem.Phys.,14(24):13377–13390.

Roberts,G., Wooster,M.J., Xu,W., Freeborn,P.H., Morcrette,J.-J., Jones,L., Benedetti,A., andKaiser,J.:LSASAFMeteosatFRPProducts:Part2–EvaluationanddemonstrationofuseintheCopernicus AtmosphereMonitoring Service (CAMS), Atmos. Chem. Phys. Discuss., 15, 15909-15976,doi:10.5194/acpd-15-15909-2015.

vanderLaan-Luijkx,I.,vanderVelde,I.,Krol,M.,Gatti,L.,Domingues,L.,Correia,C.,Miller,J.,Gloor,M., van Leeuwen, T., Kaiser, J.,Wiedinmyer, C., Basu, S., Clerbaux, C., andPeters,W. (2015).Responseoftheamazoncarbonbalancetothe2010droughtderivedwithcarbontrackersouthamerica.GlobalBiogeochemicalCycles.2014GB005082.Inpress.

Paugam,R., Wooster,M., Freitas,S.R., and ValMartin,M.: A review of approaches to estimatewildfireplumeinjectionheightwithinlargescaleatmosphericchemicaltransportmodels–Part1,Atmos.Chem.Phys.Discuss.,15,9767-9813,doi:10.5194/acpd-15-9767-2015,2015.

Paugam, R., Wooster, M., Atherton, J., Freitas, S. R., Schultz, M. G., and Kaiser, J. W. (2015).Development and optimization of awildfire plume risemodel based on remote sensing datainputs-part2.AtmosphericChemistryandPhysicsDiscussions,15(6):9815–9895.

Spessa, A. C., Field, R. D., Pappenberger, F., Langner, A., Englhart, S., Weber, U., Stockdale, T.,Siegert, F., Kaiser, J.W., andMoore, J. (2015). Seasonal forecasting of fire over Kalimantan,Indonesia.NaturalHazardsandEarthSystemSciences,15:429–442.

Tsigaridis,K.,Daskalakis,N.,Kanakidou,M.,Adams,P.J.,Artaxo,P.,Bahadur,R.,Balkanski,Y.,Bauer,S.E.,Bellouin,N.,Benedetti,A.,Bergman,T.,Berntsen,T.K.,Beukes,J.P.,Bian,H.,Carslaw,K.S., Chin,M., Curci,G.,Diehl, T., Easter, R. C.,Ghan, S. J.,Gong, S. L.,Hodzic,A.,Hoyle, C. R.,Iversen,T.,Jathar,S.,Jimenez,J.L.,Kaiser,J.W.,Kirkevag,A.,Koch,D.,Kokkola,H.,Lee,Y.H.,Lin,G., Liu, X., Luo,G.,Ma, X.,Mann,G.W.,Mihalopoulos,N.,Morcrette, J.-J.,Muller, J.-F.,Myhre,G.,Myriokefalitakis, S.,Ng,N. L.,O’Donnell,D.,Penner, J. E.,Pozzoli, L.,Pringle,K. J.,Russell,L.M.,Schulz,M.,Sciare,J.,Seland,Ø.,Shindell,D.T.,Sillman,S.,Skeie,R.B.,Spracklen,D.,Stavrakou,T.,Steenrod,S.D.,Takemura,T.,Tiitta,P.,Tilmes,S.,Tost,H.,vanNoije,T.,vanZyl,P.G.,vonSalzen,K.,Yu,F.,Wang,Z.,Wang,Z.,Zaveri,R.A.,Zhang,H.,Zhang,K.,Zhang,Q.,andZhang,X.(2014).TheAeroComevaluationandintercomparisonoforganicaerosolinglobalmodels.Atmos.Chem.Phys.,14(19):10845–10895.

Veira,A.,Kloster,S.,Schutgens,N.A.J.,andKaiser,J.W.(2015).Fireemissionheightsintheclimatesystem-part2: Impactontransport,blackcarbonconcentrationsandradiation.AtmosphericChemistryandPhysics,15(13):7173–7193.

Whitburn,S.,VanDamme,M.,Kaiser,J.,VanDerWerf,G.,Turquety,S.,Hurt-mans,D.,Clarisse,L.,Clerbaux,C.,andCoheur,P.-F.(2015).Ammoniaemissionsintropicalbiomassburningregions:comparison between satellite-derived emissions and bottom-up fire inventories. AtmosphericEnvironment,pageinpress.

Wooster,M.J.,Roberts,G.,Freeborn,P.H.,Xu,W.,Govaerts,Y.,Beeby,R.,He,J.,Lattanzio,A.,andMullen,R.:MeteosatSEVIRIFireRadiativePower(FRP)productsfromtheLandSurfaceAnalysisSatellite Applications Facility (LSA SAF) – Part 1: Algorithms, product contents and analysis,Atmos.Chem.Phys.Discuss.,15,15831-15907,doi:10.5194/acpd-15-15831-2015.

Page 53: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page53of148

5.Greenhousegases(GHG)

TheGHGsubprojectofMACC-IIImanagedandconsolidatedtheGHGprocessingchaindesignedandbuilt in the course of GEMS, MACC and MACC-II. This pioneer service, now brought to a pre-operational level, goes from series of atmospheric observations of CH4 and CO2 to global 4Dconcentration fields and 3D flux fields of these species and of N2O. It has been relying (i) on theECMWF 12-h 4D-Var to analyse theGHG concentrations in an optimalway, and (ii) on two long-window (from months to decades) variational inversion schemes to extend the analysis to thesurfacefluxes.Thistwo-tierapproachallowsthe4Dconcentrationstobeanalysedatmuchhigherglobal spatial resolution (<2,000km2) than the fluxes (>10,000km2). Further it guarantees someconsistency with the analysis of the other species within the broad MACC-III service. The GHGprocessingchainalsoincludeshigh-resolutionmedium-rangeforecastsofCH4andCO2.

5.1Observations

TheGHGprocessingchainexploitsalargevarietyofobservationsfromtheveryaccuratebutsparsesurfacemeasurementstotheglobalsatelliteobservationsoftheverticalcolumn.Therequirementson accuracy and precision forGHGmeasurements aremuchmore stringent than for species thathave not accumulated in the atmosphere: by default these data are therefore available only inresearchmode,i.e.monthsoryearsaftertheacquisitionoftherawdata.TheMACC-IIIservicehaspioneeredamuchfasterprocessinginordertomonitortheGHGsclosertoNRT.Thisfastpacehasbeenallowedbyamajoreffort fromthethreeretrieval teamstoacquireandprocessthesatellitelevel1Bdata,andthentoquality-controlanddelivertheretrievals, inquasiNearRealTime(NRT).ThisimpliedamuchhigherlevelofautomationthaninMACC-II.

IUP-UBhasdelivered retrievalsofCO2column-averageddryairmole fraction (XCO2) fromTANSO-FTS (GOSAT)within4daysafterdataacquisition.Thedeliveryhas included theaveragingkernels,prior profiles and error statistics, in addition to the XCO2 retrievals. The algorithm used for theevaluationoftheTANSOmeasurementsreliesonIUP-UB'sfull-physicsretrievalalgorithm“BESD”.Inorder to avoid data gaps after the MACC-II project, April to August 2014 data have also beendelivered in addition to the contractually-agreed delivery. The IUP-UB quasi-NRT system, whichincludesasystemforacquiringtheneededinputdata(GOSATL1BdataviaESA'sthirdpartymissionarchiveandmeteorologicaldata fromECMWF)and theprocessingof theL1BdatabyusingBESD,hasbeencontinuouslymonitoredandallneededequipmenthasbeenmaintained.Interruptionsintheproductionchainhavebeenimmediatelyresolvedasgoodaspossible.Thelongestinterruptionoccurredduring the TANSO-FTS instrumentpointingmechanism changebeginningof 2015. In thecourse of the project, theNRT system has been optimized so that stability and processing speedhave been improved. The current status of the near-real-time processor can be checked on theGOSAT BESD website (http://www.iup.uni-bremen.de/~heymann/besd_gosat.php). In addition tothecurrentstatus,globalmapsofGOSATBESDXCO2showingthespatialdatacoverageandmodelresults for comparisons can be found on that website. The quality of the data product has beenassessedbyacomparisonwithTCCONandmodeldata.Figure5.1showsanXCO2scatterplotofallcollocationsofGOSATBESDandTCCONXCO2inthetimeperiod2010to2013.Thegoodqualityofthe dataset is shownby the large correlation coefficient of 0.84 and the small scatter of about 2ppm. In a more detailed comparison, a station-to-station bias deviation of 0.43 ppm and anestimated precision (random error) of 2.1 ppm has been found (collocation criteria: 10°x10°

Page 54: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page54of148

latitude/longitude boxes around 11 TCCON sites and by using all TCCONmeasurements within 2hours before and after a GOSATmeasurement)which also confirms the good quality of the dataproduct.TheseresultshavebeenpublishedbyHeymannetal.(2015).

Figure5.1.ScatterplotofindividualGOSATBESDvs.TCCONXCO2measurementsatthelistedTCCONsites(January2010–December2013).FordetailsseeHeymannetal.,2015.

Similarly,SRONgeneratedanddeliveredretrievalsoftheCH4column-averageddryairmolefraction(XCH4)usingthe“proxy”methodwithalgorithmRemoTeCv2.0.Largeadjustmentsintheprocessingweredonetofacilitatethisgoal.AuxiliarymeteorologicalinformationwasprovidedtothealgorithmbytheECMWFoperationalanalysisinsteadofERA-Interim.TheRemoTeCalgorithmusestheproxymethod toaccount for light-pathmodifications.To thisend,XCH4 is retrievedsimultaneouslywithXCO2, neglecting scattering in the atmosphere, and their ratio is multiplied with modelled XCO2

fields. For these modelled fields, data from CarbonTracker 2013 are used for historical reasons(MACC-III data could also be used). Extrapolation is necessary for 2014 onwards because ofCarbonTrackeravailability.Fortheyearsafter2013,theCarbonTrackerCO2fieldsfromthesamedayin2013areusedaddinga2ppmyearlyincrease.GOSATL1Bare(generally)availableafter3-4daysfromtheESArollingarchive.Processingtakes1day(on1core),afterwhichtheretrievalresultsareuploadedtoanftpserver,fromwhichECMWFdownloadstheresultstoconstraintheCH4forecasts.ThewholesystemstartedinAugust2014andperformedwell,howeverdelaysintheL1BavailabilityfromtheESAservercausedgapsinthedeliveryattimes.IntheprospectofafullyoperationalNRTsystem, this aspect as well as redundancies in the processing (backup servers for processing, forinstance)havetobeaddressedproperlytopreventgapsinthedatacoverage.

ThesamepaceofNRTproductionanddeliverywasfollowedforMetop-A/IASIbyCNRS-LMD:sinceMarch2015,IASICH4fieldsaredeliveredtoECMWFatD+2.Inaddition,thespatialcoverageoftheretrievalshasbeenextendedfromthetropicalregion,whichwascoveredexclusivelysofar,totheextra-tropicalregion,therebydramatically increasingthenumberofobservationsavailableperdayfor theassimilation.These twomajor improvementsare illustrated inFigure5.2,whichshowstheIASI/Metop-ACH4retrievals forthe28thof JunedeliveredtoECMWFonthe30thof June2015at8:30am.

Page 55: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page55of148

Figure5.2.Mid-troposphericCH4asretrievedfromIASIandAMSUinstrumentsflyingon-boardMetop-Aforthe28thofJune2015generatedonthe30thofJune.

For CO2 and CH4 surfacemeasurements,NRT delivery had been already achievedwithinMACC-II.ThiseffortforNRTsurfacemeasurementscontinuedthroughoutMACC-III,withthedeliverywithin24 h by the ICOS Atmospheric Thematic Centre (ATC, located at LSCE/CEA) of data from sevenpreoperational ICOS sites in Europe, Greenland and Africa (Table 5.1). Because of their highaccuracy,theICOSinsitudatahavebeenservingforindependentvalidationofthevariousMACC-IIIGHG products. However, the current 4D-Var system is not able to keep the benefit of analysisincrementsgeneratedfromtheassimilationofsurfaceinsitudatabeyondthe12-hlengthofthe4D-Var assimilation window. A weak-constraint formulation of the 4D-var, with a parametricformulationofthemodelerror,isbeingconsideredtoaddressthisissueinthefuture.

Table5.1.ListofcontributingobservatoriestoMACC-II.Towerstationsaremulti-levelsites.

Observatory Country Acronym Type SpeciesBiscarosse France BIS Mast CO2/CH4Cabauw Netherlands CBW Talltower CO2/CH4Ivittuut Greenland IVI Mast CO2/CH4Lamto Coted’Ivoire LTO Mast CO2/CH4MaceHead Ireland MHD Mast CO2/CH4Observatoire Pérennedel’Environnement France OPE Talltower CO2/CH4/CO/N2O

Puijo Finlande PUJ Mast CO2/CH4Puy-de-dome France PUY Mast CO2/CH4

5.2Designoftheanalysis/forecastingsystemforGHGs

Despite the NRT delivery of satellite XCO2 and XCH4 retrievals, ECMWF has maintained theproductionofacyclicforecast,i.e.aC-IFSsimulationunconstrainedbytheXCO2andXCH4retrievals(Agustí-Panaredaetal.2014).Indeed,itwasfoundthattheskilloftheCO2/CH4forecastverymuchdependsontheskilloftheunderlyingmeteorologicalfieldsandthereforedegradesrapidlywiththe

Page 56: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page56of148

forecastrange.Ausefulforecaststartedfromtheanalysiswouldneedananalysisevenmorerecentthan5daysbehindthedate.Alternatively,theCO2/CH4forecastcouldberunfromtheCO2/CH4analysisusingtheanalysedmeteorologicalfieldsuntilitreachesrealtimeandthentouseforecastedmeteorologicalfields.Thisnewconfigurationcouldnotbeimplementedwithinthetimeframeoftheproject, but will be in the near future. In the meantime, it was decided to deliver both a cyclicforecast product and a 4D-Var analysis product (constrained by XCO2 and XCH4 retrievals). TheoperationalCH4/CO2analyses(Massartetal.2014)arerun5daysbehindrealtimeinsteadoftheprevious delay of 6 months for the delayed-mode analysis. This new analysis system starts eachmorningat7:00UTC.Thenumberofverticallevelswasincreasedfrom60to137andthehorizontalresolutionwaskeptat80km.TheoperationalCH4/CO2forecastrunseverydayfrom00UTCinacyclic mode. This means that the meteorological parameters are initialized using the ECMWFoperationalanalysis,buttheCO2andCH4fieldsarecycledfromtheprevious1-dayforecast.Itisrunatamuchhigherresolutionthantheanalysis,currently16kmand137levels, i.e.thesamespatialresolutionastheweatherforecast.

Theprior fluxes forCO2andCH4wereconsistentwithwhatwassetupwithinMACC-II,butanewbiogenic flux adjustment scheme (BFAS) has been implemented in C-IFS in order to correct thebackgroundCO2 biases from themodel. BFAS is nowoperational in theCO2 analysis and forecastsystems.ResultsshowthatBFASvastlyreducestheatmosphericCO2biases.Moreover,itimprovesthe synoptic variability over continental regions directly influenced by land ecosystem fluxes. Theschemederivesre-scalingfactorsforthemodelledecosystemfluxesbyusinga10-yearclimatologyof theMACCCO2optimized fluxesasa reference.Thisclimatology isadjustedwithan interannualvariability factor based on diagnostics from the operational Ensemble Prediction System (EPS) re-forecast at ECMWF. The background error statistics for the 4D-Var were also revisited using anensemblemethod.

Another improvement came from a new tracermass fixer thatwas implemented in the IFS. Thistracermass fixer ismore suitable for high resolution than the previous proportional tracermassfixer.Anevaluationwithtotalcolumnobservationsshowedthatthenewmassfixerleadstoabetterinter-hemisphericgradientofCO2andCH4inthemodel.

Ashort re-analysiswasperformedfor2010toevaluate theupgradepackagebrought in theCO2/CH4analysis/forecastsystem.ItshowsconsiderableimprovementofthequalityoftheMACC-IIICO2/CH4products(Figure5.3).Morerecently,firstXCO2retrievalsfromOCO-2werereleasedbyNASAandtheyhavebeenmonitoredoveraperiodofafewmonths.

Page 57: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page57of148

Figure5.3.DailymeanCO2drymolarfraction(ppm)fromNOAA/ESRLobservationsattheParkFallstalltower(Wisconsin,USA)inMarch2010anddifferentcyclicforecastexperiments:withBFAS(cyan),withoutBFAS(red),withMACCoptimizedCO2fluxes(green)andwitha10-yearclimatologyoftheMACCoptimizedCO2fluxes(blue).

ThequalityofMACCCO2forecastsandanalysisdatasetshasbeenevaluatedbyECMWFusingTCCONdata(Figure5.4).Differentleadtimesintheshortrange(day-1)andmediumrange(day-4andday-10)havebeenconsidered.Theday-1forecastskill isgenerallyhigh inthewinterwithameanbiasand standard deviation of -0.42 ppmand 0.79 ppm respectively. In the summer themeanbias islarger,withvaluesof-1.21ppm,andastandarddeviationof0.75ppm.Theforecasterrorsincreasewith forecast lead time, as expected. This is particularly pronounced in the summer, when thestandarddeviation(ormeanscatter)increasesto1.37ppm.

Figure5.4.TimeseriesofXCO2(inppm)overtheTCCONstationsofLauder,LamontandSodankylä.Blackdots:TCCONmeasurements.Cyanline:XCO2fromtheanalysesmadewiththetime-evolvingMACCsystem.Cyandots:XCO2fromtheseanalysesusingtheTCCONaveragingkernels.Redline:XCO2fromthereanalysis(newMACCsystem).Reddots:XCO2fromthere-analysisusingtheTCCONaveragingkernels.ThefigureclearlyshowstheimprovingqualityoftheMACCanalysissystemovertime.

ThequalityofMACCCH4forecastsandanalysisdatasets,andofcorrespondingXCH4retrievalsfromGOSAT,wasevaluatedforJanuaryandAugust2014byULEICandECMWF.TheaimofthisworkwastodeterminethebiasesandvariabilityoftheMACCandsatellitedatasetsbyinter-comparisonwithCH4measurements fromtheTCCONground-basedmeasurementnetwork.FortheMACCdatasets,

Page 58: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page58of148

bothlowandhighresolutionforecastsweretestedtoprovideinformationofbiasesforwinterandsummerXCH4concentrationsandforcorrespondingforecaststhatwereproducedwitha4-dayand10-dayleadtime.Similarly,NRTXCH4GOSATretrievalsproducedbySRONandaMACCanalysisthatassimilated the SRONXCH4 datawere tested (although this investigationwas limited due to dataavailability issues). The inter-comparison showed that the NRT XCH4 GOSAT retrievals from SRONwere negatively biasedwith respect to TCCON and that for some stations the uncertaintieswerelarge. The MACC analysis that assimilated the satellite retrievals were also biased low, therebyhighlightingthepropagationofsatelliteuncertaintiesintotheresultingMACCanalysis.ThequalityoftheMACCforecastsvarieswiththemonth,thelocation,themodelresolutionandtheforecastleadtime(Figure5.5).NotethatTCCONstationsarenotalwaysrepresentedforallseasonsandthereforeaveragedbiasanduncertaintyestimatesshouldbeusedwithcaution.Generallyitwasfoundthatforsomeof the TCCON stations, the correlation reduced and the uncertainties increased (more than95% significant) for forecastswith lead timesof 10days as expected. Thehigh resolutiondatasettendedtoshowlargerbiasescomparedtothelowresolutionforecasts.Someofthecausesoferrorswerehypothesisedtobeduetotheintensityoffireandanthropogenicsurfacefluxesbeingstrongerin the high resolution forecasts combined with the errors in the transport pathways. Furthercontrolled studies should be conducted to determine the impact of these factors on the biasesobserved.

Figure5.5.Time-seriesoftotalcolumnmethane(ppb)forLamont(36.60°N,97.486°W)inJanuary(J–F)andAugust(A–S)withthedailyaveragedTCCON(darkgreencircles)anddailyaveragedMACC-IIIforecastsasafunctionofforcastleadtime(1-day:red,4-day:purpleand10-day:blue).Theaveragedmonthlydifference,standarddeviationandcorrelationcoefficientsarealsohighlightedtoindicatethequalityofMACC-IIIXCH4againstinsitumeasurementsfromTCCON.Thefiguredemonstratesthatlongerleadtimesresultinlargerbiasesandlargervariability.

5.3Surfacefluxinversions

5.3.1CH4

The series of 6-monthly “delayed-mode” global CH4 flux inversions assimilating satellite XCH4

retrievals has been extended for the period from07/2013until 06/2014by EC-JRC and TNO. Thenew “delayed-Mode” CH4 inversions show similar regional emission patterns as for the previousperiods,particularlyovertheUnitedStatesandTropicalAfrica.Furthermore,adetailedcomparisonstudy has been performed using different XCH4 satellite products from SCIAMACHY and GOSAT,includingadditionalXCH4productsdevelopedwithintheESAGHGclimatechange initiativeproject(Alexe et al. 2015; Figure 5.6). This study showed qualitatively consistent regional CH4 emissionpatterns(althoughthemagnitudeoftheemissionsderivedfordifferentTRANSCOMregionsdifferbyup to ~10-15 Tg CH4/yr). TheMACC-III “delayed-Mode” CH4 inversions have been validated using

Page 59: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page59of148

various independent observation data sets. The validation versus the NOAA ship and aircraftmeasurementsshowthat ingeneraltheremotefreetroposphere isrealisticallyreproducedbythe3Dmodelfields.However,thecomparisonsvs.theFourierTransformSpectrometers(FTS)fromtheTotalCarbonColumnObservingNetwork(TCCON)showsomelatitudedependentbias,whichcouldpointtosomeshortcomingsofthemodeltosimulatethestratosphere,especiallyathighlatitudes.Onthetechnicalside,amajorefforthasbeenundertakenwiththefurtherdevelopmentofthenewTM5-4DVAR “pyshell” versionwith enhancedmodularity,merging various specific features of theJRCTM5-4DVARversion(“T38”)intotheTM5-4DVARpyshellversiondevelopedbytheinternationalTM5community.ThismergedversionwillfacilitatesignificantlythefurtherdevelopmentwithintheinternationalTM5-4DVARcommunity.

Figure5.6.ComparisonofderivedglobalCH4emissionsusingGOSATXCH4 retrievals (GOSAT-SRON-PX:RemoteCPROXYv2.0,usedinMACC)andusingSCIAMACHYretrievals(IMAPv55)(Alexeetal.,2015);thefigureshows2010-2011averageemissions(left)andthedifferencecomparedtoareferenceinversionusingonlyNOAAsurfaceobservations(right).

5.3.2CO2

TheCO2 flux inversion assimilates surface air-samplemeasurements only. It has beenupdated byLSCEinMay2015(v14r1),therebycontinuingthepaceoftworeleasesperyearthathasbeenfirstachievedin2011.Version14.1 improvesonversion13.1(delivered inthe lastdaysofMACC-II)onseveralaspectsdescribedbelow.It isthefirstreleasetocover2014,butitonlystartsin2006.Thereason for not having processed the years before 2006 came from the technical nature of thisrelease. Indeed it includesachangeoftheLMDZtransportmodelconfigurationthathadnotbeentestedonmulti-yearCO2 inversionsbefore:processing9years(2006-2014)ratherthan36allowedtesting the new configurationwith a reasonable computational cost (still a fewweeks on a high-performance-computing system). Thenext release (v14.2, generated after the endof the project)covers theperiod1979-2014. The change in the LMDZ transportmodel consisted in replacing thedeep convectionmodel of Tiedtke (1989) by that of Emanuel (1991). This configuration of LMDZcorresponds to the one (LMDz5A) developed and used for the fifth phase of the CoupledModelInter-comparisonProject(CMIP5),Forthesimulationoftracers,themainimpactisaslightlyslowerinter-hemispheric exchange, but at the cost of a doubling of the computing time (Locatelli et al.,

Page 60: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page60of148

2015).v14r1alsoincludedarevisionoftheassimilatedobservationswith14measurementrecordsaddedandtworemoved.Theevaluationofbothv13r1andv14r1showsthat theyperformbetterthan a baseline inversion that just assimilates the annual global growth rate. They fit theirassimilateddatawithintheassignedstandarddeviationoftheobservationuncertainty.IntermsofRMS,theinversionsusuallyfittheirassimilateddata,thecolumnmeasurementsandtheaircraftfreetropospheremeasurementsbetterthan2ppm(themedianoftheRMSisusuallyabout1ppm).TheRMSfitwithaircraftprofilesintheboundarylayerisusuallybetterthan3ppm.Differencesbetweenthe two flux products are usually within 1 sigma of the posterior uncertainty (Figure 5.7), butchanges in Northern Africa and Temperate Eurasiamakes v14r1more consistentwith bottom-upresultsthanv13r1.

TheassimilationofGOSAT/TANSOXCO2retrievalshasbeenfurthertested,andinconsistencieshavebeen found between the retrieval schemes and the inversion schemes, in terms of statisticalassumptions (Chevallier 2015). These inconsistencies may explain some of the surprising CO2

atmospheric inversion results obtained with the existing GOSAT retrieval products and motivatefurtherinvestigation.

Figure5.7.InferrednaturalCO2annualflux(withoutfossilfuelemissions)averagedovertheTransCom3landregionsandovertheglobe.Thebluecurvecorrespondstov13r1withits1-sigmauncertaintyandtheredonecorrespondstov14r1.Inthesignconvention,positivefluxescorrespondtoanetcarbonsourceintotheatmosphere.

5.3.3N2O

TheN2OfluxinversionhasbeenupdatedbyNILUinFebruary2015(v12r1).Itcoversthe1996-2012period and assimilated concentration measurements from 124 ground-based stations, ship andaircraftcampaigns.Theoptimizedfluxeswerevalidatedbycomparingtheposteriorforwardmodelsimulationswithindependentobservations,namelythosefromtheHiaperPole-to-PoleObservation(HIPPO)campaigns.Theposteriorfluxesprovidedabetterfittotheobservationscomparedtotheprior. Furthermore, the posterior model was in closer agreement with the observed globaltroposphericgrowthrate,of0.81ppb/y,thanthepriormodel,whichhadatoohighgrowthrate,of

Page 61: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page61of148

1.23ppb/y.Accordingly,themeanposteriorglobalsource,17TgN/ywaslessthanthatoftheprior,19 TgN/y. Considerable variability was found in the global source from year-to-year, from aminimumof15.5TgN/yin2007toamaximumof19.7TgN/yin2010.However,nosignificanttrendin the global sourcewas found over the period 1996 to 2012. This is in contrast to a number ofcontinental regions,whichdidshow important trends. InSouthandEastAsia,SouthAmerica,andAfrica, increasing trends in N2O emissions were found, consistent with increases in N-fertilizerconsumption. In contrast, negative emission trends were found in North America and Europe. InEurope, this may be explained by the reduction in N-fertilizer consumption, whereas in NorthAmerica, N-fertilizer consumption has slightly increased indicating that the emission reduction islikely rather due to climatic changes, such as soil moisture, N-deposition, and/or agriculturalpractices.

5.4References

Agustí-Panareda,A.,S.Massart,F.Chevallier,S.Boussetta,G.Balsamo,A.Beljaars,P.Ciais,N.M.Deutscher,R.Engelen,L.Jones,R.Kivi,J.-D.Paris,V.-H.Peuch,V.Sherlock,A.T.Vermeulen,P.O.Wennberg,andD.Wunch,2014:ForecastingglobalatmosphericCO2.Atmos.Chem.Phys.Discuss.,14,13909-13962,doi:10.5194/acpd-14-13909-2014

Alexe,M.,P.Bergamaschi,A.Segers,R.Detmers,A.Butz,O.Hasekamp,S.Guerlet,R.Parker,H.Boesch,C.Frankenberg,R.A.Scheepmaker,E.Dlugokencky,C.Sweeney,S.C.WofsyandE.A.Kort,InversemodelingofCH4emissionsfor2010–2011usingdifferentsatelliteretrievalproductsfromGOSATandSCIAMACHY,Atmos.Chem.Phys.,15,113-133,2015.

Chevallier,F.:OnthestatisticaloptimalityofCO2atmosphericinversionsassimilatingCO2columnretrievals,Atmos.Chem.Phys.Discuss.,15,11889-11923,doi:10.5194/acpd-15-11889-2015,2015.

Emanuel,K.:ASchemeforRepresentingCumulusConvectioninLarge-ScaleModels,J.Atmos.Sci.,48,2313–2329,doi:10.1175/1520-0469(1991).

Heymann,J.,Reuter,M.,Hilker,M.,Buchwitz,M.,Schneising,O.,Bovensmann,H.,Burrows,J.P.,Kuze,A.,Suto,H.,Deutscher,N.M.,Dubey,M.K.,Griffith,D.W.T.,Hase,F.,Kawakami,S.,Kivi,R.,Morino,I.,Petri,C.,Roehl,C.,Schneider,M.,Sherlock,V.,Sussmann,R.,Velazco,V.A.,Warneke,T.,andWunch,D.:ConsistentsatelliteXCO2retrievalsfromSCIAMACHYandGOSATusingtheBESDalgorithm,Atmos.Meas.Tech.,8,2961-2980,doi:10.5194/amt-8-2961-2015,2015.

Locatelli,R.,Bousquet,P.,Hourdin,F.,Saunois,M.,Cozic,A.,Couvreux,F.,Grandpeix,J.-Y.,Lefebvre,M.-P.,Rio,C.,Bergamaschi,P.,Chambers,S.D.,Karstens,U.,Kazan,V.,vanderLaan,S.,Meijer,H.A.J.,Moncrieff,J.,Ramonet,M.,Scheeren,H.A.,Schlosser,C.,Schmidt,M.,Vermeulen,A.,andWilliams,A.G.:AtmospherictransportandchemistryoftracegasesinLMDz5B:evaluationandimplicationsforinversemodelling,Geosci.ModelDev.,8,129-150,doi:10.5194/gmd-8-129-2015,2015.

Massart,S.,A.Agustí-Panareda,I.Aben,A.Butz,F.Chevallier,C.Crevoisier,R.Engelen,C.Frankenberg,andO.Hasekamp,2014:AssimilationofatmosphericmethaneproductsintheMACC-IIsystem:fromSCIAMACHYtoTANSOandIASI.Atmos.Chem.Phys.,14,6139-6158,doi:10.5194/acp-14-6139-2014

Page 62: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page62of148

Tiedtke,M.:Acomprehensivemassfluxschemeforcumulusparameterizationinlarge-scalemodels,Mon.WeatherRev.,117,1779–1800,doi:10.1175/1520-0493(1989).

Page 63: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page63of148

6.GRG:GlobalReactiveGases

The main objective of GRG in MACC-III was to “provide continuation of the global atmosphericcompositionservicesforthestratosphereandtroposphereusingthenewlydevelopedC-IFSmodelsystem.”ThedevelopmentofC-IFS,i.e.havingthechemistrymodulesfullyintegratedwithintheIFSmodel framework instead of using coupling software to exchange information between twoindependentmodels,startedduringtheMACCprojectandhasnowledtothreeC-IFSconfigurationsthatarepracticallyreadyforoperationaluse.Indeed,fromSeptember2014onward,theC-IFS-CB05model the chemistry of which is based on the TM5 chemistry transport model superseded thepreviously used coupled IFS-MOZART model for the daily analyses and forecasts of the globalreactivegasesdistributionsinthetroposphere.DuringMACC-IIIC-IFS-CB05hasbeenintegratedwiththe BASCOE stratospheric chemistry scheme. The new system C-IFS-CB05-BASCOE will allow thecontinuation of stratospheric reactive gases services in the near future. The other two modelconfigurationsareC-IFS-MOCAGEandC-IFS-MOZ.Thesetwomodelscontainchemistryschemesforthetroposphereandstratosphere.Bothmodelsaretechnicallyrunningandhaveundergone initialevaluation.Weexpectthemtobereadyforoperationalusesoon.

WiththemovetoC-IFStheglobalreactivegasesanalysesandforecastshavebecomemoreefficientand more sustainable, because any IFS cycle upgrades or changes of ECMWF supercomputinghardware is greatly facilitated if the code is integrated in onemodel framework. The changewasannouncedearlytotheusercommunityandtherewerenocomplaintsaboutdegradationofservicequality or changes in themodel output data distribution. It was noted that the absence of a fullstratospheric chemistry scheme in C-IFS-CB05 creates an issue with data users from the satellitecommunitywho have begun to useMACC forecasts directly in their retrieval algorithms. Yet, theswitchwasunavoidable,becausetheoldcoupledmodelset-upcouldnotbeportedontothenewcomputerhardwarewhichECMWFacquiredatthebeginningofMACC-III.WiththereplacementofC-IFS-CB05 by C-IFS-CB05-BASCOE suitable C-IFS products for stratospheric composition shouldbecome available again soon. In any case, stratospheric ozone in the C-IFS-CB05 system isconstrained from assimilation of ozone analyses, while the “external” stratospheric servicesprovidedbytheBASCOEandSACADAdataassimilationsystemswerecontinuedduringMACC-IIIandtheresultsweredisplayedandmadeavailablethroughthestratosphericservicewebpagesatBIRA(http://www.copernicus-stratosphere.eu).TheywereusedextensivelybytheWMOozonebulletins(e.g.http://www.wmo.int/pages/prog/arep/WMOAntarcticOzoneBulletins2015.html).

In preparation for the operational Copernicus Atmosphere Monitoring Service, the GRG teamdevelopedaservicestrategy,whichinvolvestheparalleloperationofthreechemistrymodulesinC-IFS(oneofthemindataassimilationandforecastmode,theothertwoinforecastmodeonly).Theserunswouldbecomplementedbyhigh-resolution tracer forecast runs in supportof theanalysisoflong-rangetransporteventsandforplanningofscientific fieldcampaigns.Allglobal reactivegasesservices would then be provided within the C-IFS framework and therefore be based on oneconsistentsetofphysicalparameterisations.

6.1Reactivegasesserviceproducts

Delivery of tropospheric reactive gases products continued without major changes. The centralMACCwebpagesatECMWFdisplaystandardmapsofthedailyanalysesandforecasts(upto5days)atthesurface,850hPa,500hPa,300hPa,and50hPaofozone,carbonmonoxide,nitrogendioxide,

Page 64: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page64of148

formaldehyde,andsulphurdioxide.Comparedtothe IFS-MOZsystemthatwasusedduringMACCandMACC-II,theforecastresolutionincreasedfromabout110km⋅110kmto80km⋅80km.ForuserswithanaccountatECMWF,theanalysesandforecastsareavailabledirectlyfromECMWFservers.All other users can retrieve these data products (and several other reactive gases and aerosolparameters)fromthewebcoverageserviceatJülich(http://join.iek.fz-juelich.de/macc/access).Thisservicealsoofferssomesimplevisualisationoptions.Script-basedaccessispossiblethrough:

http://ows-server.iek.fz-juelich.de/MACC_g4e2_fc-3hourly_ModelLevel?service=WCS&acceptversions=1.1.2&Request=GetCapabilities&sections=Contents.

ThetroposphericanalyseswereregularlysubjectedtovalidationbytheVALproject.IngeneralthesimulatedconcentrationsofozoneandCOagreedwellwithobservationsfromroutineaircraftandsurfacestationmeasurements.However,theglobalsystemexhibitsahighbiasofsurfaceozoneinthenorthernmidlatitudesinsummer,andahighbiasofCOinthetropicsandsouthernhemisphereyear round (Eskes et al., 2015b).Asnotedpreviously, assimilationof satellite data aloneprovidesonly poor constrains on trace gas concentrations near the surface. It should therefore beinvestigated if it ispossibletomakeuseofglobalsurfaceobservationsforthedataassimilationaswell.Figure6.1showsanexampleofthecomparisonbetweentheoperationalMACC-IIIC-IFS-CB05modelandsurfaceozoneobservationsfromtheEMEPstationSandveandtheGermanUBAstationWesterland,respectively.

Figure6.1.ObservedandsimulatedsurfaceozonemixingratiosfromSandve,Norway(top)andWesterland,Germany(bottom)fortheperiod1stSeptember2015to31stAugust2015.TheobservationaldatawereobtainedfromtheTOARsurfaceozonedatabaseinJülich.

Page 65: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page65of148

The stratospheric services continued to display and document the NRT analyses of stratosphericcomposition delivered by C-IFS-CB05 (ozone only), BASCOE, SACADA and TM3DAM (total ozonecolumnsonly)onadedicatedwebsitewhichisdevelopedandhostedbyBIRA-IASB.Thissitereceivesat least300uniquevisitorspermonth,andregisterspeaksofupto700uniquevisitorsduringthemonthsofSeptember, i.e. theAntarcticozoneholeseason.Theanalyses,aswellas somederivedproducts,wereusedextensivelybytheWMOozonebulletins1.

An innovativestratosphericservicestartedduringMACC-III is thequick-lookviewof theensembleforecastsofthepolarvortex.Theseviewsconsistofmosaicsofmapsshowingthe9-dayand15-dayforecastsofPotentialVorticity(PV)inthelowerstratosphere,wherepolarozonedepletionhappens.These maps are derived from a subset of the ECMWF Medium Range Ensemble Forecast. Theprimary intention is to allow a subjective evaluation of the probability of vortex splits during thepolarspring,becausesuchvortexsplitsmarktheendofozoneholeevents.

6.2C-IFSdevelopmentduringMACC-III

While the major tasks of implementing TM5-CB05, MOCAGE, and MOZART into C-IFS weretechnically accomplished already during MACC-II, a lot of progress was made on all three C-IFSconfigurationsduringMACC-III.BesidestheadaptationofcodetoanewIFScycleandtestingonthenewhardwareofECMWF,wewouldliketohighlightthefollowingdevelopments:

A major accomplishment during the MACC-III project was the extension of the C-IFS-CB05 withstratosphericchemistryoriginatingfromtheBASCOEsystem.Thischemicalschemeconsistsof142gas-phase reactions, 9 heterogeneous and 52 photolytic reactions, and poses quite differentrequirements to the model than what was needed in the troposphere. For instance, theparameterizationofphotolysisratesinthestratosphererequiresshort-waveradiation,andneedstobe computed up to solar zenith angles of 96°, while the effects of tropospheric aerosol can beneglected.Alsohalogenchemistry,aswellasanexplicitdescriptionofextendedradicalchemistry,caninturnbeneglectedinthetroposphere.

Thereforeitwasdecidedtokeepthetwoschemesside-by-side,ratherthantomergethemintoone.Now either tropospheric or stratospheric chemistry is switched on at every single grid location,dependingonitslocationwithrespecttothetropopause.

This approach has proven toworkwell in terms of chemical composition in the troposphere andstratosphere,withoutstrongspuriousgradientsinthetracegascompositionattheinterface.Polarozonedepletionduetoheterogeneouschemistryhasbeensuccessfully implemented,asshown inFigure 6.2 by the excellent agreement between the new model configuration and observations.Nevertheless,somelong-livedspeciesexhibitbiasesthatgrowovertimescalesofseasonstoyears.Thismayexplainthedisagreement,whichcanbeseeninFigure6.3above30hPa.Workisongoingtoresolvesuchissues.

1E.g.http://www.wmo.int/pages/prog/arep/WMOAntarcticOzoneBulletins2015.html

Page 66: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page66of148

Figure6.2.EvaluationoftroposphericozoneagainstWOUDCozonesondesatSyowastationduringSeptember-December2009,i.e.aftera1.5yearfreemodelrun.Black:ozonesonde,Red:CIFS-CB05-BASCOE,blue:CIFS-CB05.

TheMOCAGE chemical schemenow includes anupdated cycle for sulphur compounds, aswell asimprovements in the treatment of stratospheric nitrogen compounds. The MOCAGE chemicalmodulesthatareinterfacedtoC-IFShavebeensynchronizedwiththeMOCAGE-CTMdevelopmentsthrough a common interface library (SUGAR). First C-IFS-MOCAGE runs have been performedsuccessfully. The results have been compared toMOCAGE-CTM runs, showing a good agreementbetweenbothmodels.Finally,atestruninassimilationmodehasbeenperformedsuccessfully.

AstableandreliableversionofCIFS-MOZcouldbeestablishedin2014.BytheendofMACC-III,theCIFS-MOZhasbeenupdatedtoIFScycle41R1andseveraltestsimulationsofupto1yeardurationhavebeenperformed.ResultsfromthefirsttestsimulationswerepresentedattheMACC-IIIGeneralAssembly in Reading, February 2015. At that time, C-IFS-MOZ was able to simulate globalatmospheric composition ina reasonablemanner, comparable to theotherCTM implementationsand to MOZART standalone simulations. The results indicate a slight overestimation of NHtroposphericozone,whichincreasestowardsthetropopause.AsintheoriginalMOZART-3CTM,themodeldoesnotalwaysproducestratosphericozonemaximaandminimaattherightaltitudes.Wehope to overcome this with the implementation of a new stratospheric PSC scheme. A detaileddescriptionofthegas-phasemechanism,photolysisratesandstratosphericheterogeneousreactionsforCIFS-MOZisgiveninSteinetal.,2015.

6.3Progresswithreactivegasesdataassimilation

C-IFS-CB05 works well in data assimilation mode and has been running stably in NRT sinceSeptember 2014. Extensive comparisons of a C-IFS-CB05 data assimilation run for the year 2008(assimilatingozone,CO,andNO2)yieldedbetterresultsthanacontrolrunwithoutdataassimilationandtheMACC-IIreanalysis(Innessetal.2015).AssimilationtestshavebeencarriedoutwithafirstversionofC-IFS-BASCOE-CB05.Theresultsshowthat the assimilation of stratospheric profile and total column O3 retrievals can constrain thestratospheric ozone well. Despite the much greater complexity of the BASCOE stratosphericchemistryscheme,theresultingstratosphericO3analysisfieldisveryclosetothatobtainedwithanC-IFS-CBO5assimilationexperiment (see Figure6.3). Thisunderlines the finding that stratosphericozonecanbewellconstrainedfromcurrentsatelliteinstrumentation.

Page 67: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page67of148

Figure6.3.RelativedifferencesbetweenC-IFSprofilesandozonesondesoverEurope(left),theTropics(centre)andtheAntarctic(right)averagedoverMarchandApril2008.RedlinesshowprofilesfromC-IFS-BASCOE-CB05experiments,bluelinestheresultsfromC-IFS-CB05experiments.Analysisexperimentsareshownbysolidlines,thecontrolrunswithoutassimilationofO3retrievalsindottedlines.

Further data assimilationworkhas focusedonpreparingbackgrounderror statistics for increasedvertical resolution,with theNationalMeteorological Center (NMC; Parrish andDerber, 1992) andEnsemble Data Assimilation (EDA) methods, and on monitoring and experimental assimilation ofnew data sets (including OMPS O3, GOME-2 O3 profile and GOME-2 NO2 retrievals). In the EDAmethod the background errors are calculated from an ensemble of C-IFS forecast runs thatcontained 10 members with perturbations to the model physics, observations, and sea surfacetemperatures.Differencesbetweenpairsofbackgroundfieldsareusedtodeterminethestatisticalcharacteristicsofthebackgrounderrors(seeInnessetal.,2015).

AnalysisoftheimpactoftroposphericNO2assimilationinC-IFSshowedthattheeffectontheNO2fields isoverall small. Thereare several regionswhere there is largeandsystematicdisagreementbetween satellite data and the C-IFS tropospheric NO2 columns in spite of assimilation. Differentsatellitedataproductsshowgoodagreementintheseregionsindicatingthattheproblemisrelatedtotheforwardmodelratherthanthedataassimilationitself,presumablybecauseofuncertaintiesintheNO2emissions.Apparently,theshortlifetimeofNO2incombinationwiththeconstantemissions(nodiurnalorweekdaycyclesareimplementedinthemodel)overruleanyeffectoftheassimilation.

ImprovedSO2assimilationofvolcanicplumeswasobtainedafterintroductionofGOME-2dataandapplicationofa flaggingprocedure tomaskout thesubstantialbackgroundnoise fromthissensor(seereportD20.1).AssimilationofflaggeddatawiththeMACCNRTmodelwassuccessfulprovidinga better SO2 analysis field than using OMI data alone. After extensive testing the assimilation ofGOME-2SO2retrievalswasincludedintheMACCCY41r1e-suite.6.4Theuseofcasestudiestoguideservicedevelopment

During MACC-III, three case studies were performed and analyzed with the objective to identifyissues with chemical parameterisations, modeled transport of air masses, or input data.Furthermore,thesecasesstudiescanbeusedtotestnewdataproductsorcodedevelopments.

Inthefirstcasestudy,tracegasemissionandplumeevolutionduetoborealfiresinCanadawere assessed. The study yearwas 2008, becauseof the availability of an extensive fielddatasetfromtheARCTAScampaign(Jacobetal.,2010).Thestudywascarriedoutinclose

Page 68: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page68of148

collaborationwith the FIR project in order to identify potential issueswith theGFAS fireemissionsystemandwithGHGinordertolinkreactivegasesandgreenhousegasemissions.ThestudyobjectivesweretoassesstheintegralabilityoftheCAMSoperationalsystemtodescribe the emissions and chemical evolution of trace gas composition in boreal fireplumes,andtoinvestigatesensitivities,whichcanexplaindiscrepanciestoin-situtracegasobservations.ForthispurposeweanalysedtheobservationsofFireRadiativePower(FRP)for the August 2008 period and executed various sensitivity runswith theGHG andGRGsystems adopting the GFAS emissions. The availability of high-resolution CO2 and COforecastsallowed theassessmentof themodifiedcombustionefficiency,while themodelforecasts of reactive gases gave insight in the emission factors formultiple non-methanehydrocarbonsaswellasNOx.OuranalysissuggeststhattheabilityoftheMACCsystemtomodelborealfireplumesisinherentlyconstrainedbythelimitationsfromtheobservationalsystemused inGFAS,which could explainmuch of the biases seen in the comparison oftracegasconcentrationsagainstin-situobservations.Ontheotherhand,theanalysisoftheobserved MCE and EF’s showed generally good agreement to assumptions in GFAS,althoughwesuggestthattheEFforNOxisover-estimated,seealsoFigure6.4.

Figure6.4.Scatterplotsofobserved(black)andmodelled(red)NO2againstCOforafireplumeoverSaskachewan,sampledon1July2008.Left:usingtheoriginalGFASemissionfactorforNOx,andrightusinganoptimizedscalingfactor.

Thesecondstudylookedintothechemicalandmeteorologicalaspectsofthe2014Antarcticozoneholeand itstemporaldevelopment. IthighlightedthecontributionoftheMACCanalysesproducts(MACC o-suite, BASCOE and TM3DAM) to the corresponding Antarctic Ozone Bulletin2 issued byWMO/GAW,anddelivered someadditional resultsobtained inMACCw.r.t. themonitoringof theAntarcticozoneholein2014.Thecomparisonwithobservationsreliedonassimilated(AuraMLS)aswellas independent (OMPS,OSIRIS)observationsby limb-observingsatellite instruments, showingagain the excellent quality of the MACC analyses. The study also looked at ozone sondemeasurementsattheSouthPoleandatUshuaia,whichistypicallyattheedgeofthepolarvortex.The latter stationwas selected because theMACC stratospheric ozone service helped the stationoperatorstoplanandinterprettheirozonesoundingsduringthe2014ozoneholeseason.Comparedto a typical episode, in 2014 the polar vortex was shifted towards the Atlantic sector and SouthAmericaduring longperiods.This ledto lowozoneoverstationsreachingas farnorthasUshuaia.StationsfacingthePacificsectorwereoutsideofthevortexforlongperiodsandexperiencedlargetotalozonevalues.2http://www.wmo.int/pages/prog/arep/WMOAntarcticOzoneBulletins2014.html

Page 69: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page69of148

Finally,anothercasestudyinvestigatedthetrackingofvolcanicplumesthroughtheassimilationofSO2 satellite observations from the GOME-2 sensor. An analysis of two volcanic eruptions(Barbadunga,Iceland,andPicodoFogoonCapeVerde)in2014withafocusontheperformanceandpossible improvements of the SO2 data assimilation (see report D23.2) showed that a volcanicactivitydetectionalgorithmdevelopedbyDLRandBIRA(Brenotetal.2014)wasabletosuccessfullydetectandtrackthevolcanicSO2plumes.ThealgorithmwasadjustedtoidentifytheentirevolcanicSO2 plume andnot only takes into account different threshold values for the vertical SO2 columndependingontheproximitytoknownvolcanoesorpollutedareas(anthropogenicortheSAA).Thealgorithmrequiresthatacertainamountofneighbouringpixelsmustfulfilthesecriteria.

AssimilationoftheflaggedvolcanicSO2datawiththeMACCNRTmodelwassuccessfulandallowed for theestablishmentof an SO2 analysis fieldwhichwasbetter than the analysisfieldbasedonOMIdataalonewhereonlyfewOMISO2observationscouldbeassimilatedinthecaseofBardarbungaduetoOMI’srowanomalyandastrictselectioncriterion.

Figure6.5showsthetotalcolumnSO2analysisfieldsfromtheGOME-2assimilationexperimentandtheMACCNRTanalysis,aswellas theassimilatedGOME-2andOMIpixelson20September2014during the Bardarbunga (Iceland) eruption. Both instruments detect a SO2 plume that is beingtransportedsouthwardsfromIceland,butonlyasingleOMIobservationisactuallyassimilatedintheMACCNRTanalysis.TheassimilatedGOME-2pixelscover theplumesuccessfullyandtheresultingSO2 analysis field is better. After extensive testing the assimilation of GOME-2 SO2 retrievals wasincludedintheMACCCY41r1e-suite.

Figure6.5MACCvolcanicSO2forecastusingassimilatedGOME-2data(left)andOMIdata(right)on20September2014duringtheIcelandicBardarbungaeruption.Thecoloreddotsshowtheassimilateddatapointsfromthetwosensors.

6.5References

Brenot, H., Theys,N., Clarisse, L., vanGeffen, J., vanGent, J., Van Roozendael,M., van der A, R.,Hurtmans,D.,Coheur,P.-F.,Clerbaux,C.,Valks,P.,Hedelt,P.,Prata,F.,Rasson,O.,Sievers,K.,andZehner,C.:SupporttoAviationControlService(SACS):anonlineservicefornear-real-

Page 70: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page70of148

time satellitemonitoring of volcanic plumes, Nat. Hazards Earth Syst. Sci., 14, 1099-1123,doi:10.5194/nhess-14-1099-2014,2014.

Eskes,H.,Huijnen,V.,Arola,A.,Benedictow,A.,Blechschmidt,A.-M.,Botek,E.,Boucher,O.,Bouarar,I., Chabrillat, S., Cuevas, E., Engelen, R., Flentje, H., Gaudel, A., Griesfeller, J., Jones, L.,Kapsomenakis, J., Katragkou, E., Kinne, S., Langerock,B., Razinger,M., Richter,A., Schultz,M., Schulz,M., Sudarchikova,N., Thouret, V., Vrekoussis,M.,Wagner, A., and Zerefos, C.:Validationof reactivegasesandaerosols in theMACCglobalanalysisand forecast system,Geosci.ModelDev.Discuss.,8,1117-1169,doi:10.5194/gmdd-8-1117-2015,2015.

Eskes, H.J., T. Antonakaki, A. Arola, A. Benedictow, A. Blechschmidt, E. Botek, S. Chabrillat, Y.Christophe, E. Cuevas, H. Flentje, A. Gaudel, V. Huijnen, J. Kapsomenakis, S. Kinne, B.Langerok, M. Razinger, A. Richter, M. Schulz, V. Thouret, M. Vrekoussis, A. Wagner, C.Zerefos, Validation report of the MACC near-real time global atmospheric compositionservice:Systemevolutionandperformancestatistics,statusupto1June2015,MACCreport,17 June 2015, available at http://www.gmes-atmosphere.eu/services/aqac/global_verification/validation_reports/,2015b.

Flemming, J., Inness,A., Flentje,H.,Huijnen,V.,Moinat, P., Schultz,M.G., and Stein,O.: Couplingglobal chemistry transportmodels to ECMWF's integrated forecast system,Geosci.ModelDev.,2,253-265,doi:10.5194/gmd-2-253-2009,2009.

Gaudel,A.,H.Clark,V. Thouret, L. Jones,A. Inness, J. Flemming,O. Stein,V.Huijnen,H. Eskes,P.Nedelec,D.Boulanger:OntheuseofMOZAIC-IAGOSdatatoassesstheabilityoftheMACCReanalysistoreproducethedistributionofO3andCOintheUTLS,submittedtoTELLUSB,2015.

Inness, A., Blechschmidt, A.-M., Bouarar, I., Chabrillat, S., Crepulja, M., Engelen, R. J., Eskes, H.,Flemming,J.,Gaudel,A.,Hendrick,F.,Huijnen,V.,Jones,L.,Kapsomenakis,J.,Katragkou,E.,Keppens,A.,Langerock,B.,deMazière,M.,Melas,D.,Parrington,M.,Peuch,V.H.,Razinger,M., Richter, A., Schultz, M. G., Suttie, M., Thouret, V., Vrekoussis, M., Wagner, A. andZerefos, C.: Data assimilation of satellite-retrieved ozone, carbon monoxide and nitrogendioxide with ECMWF’s Composition-IFS, Atmos. Chem. Phys., 15(9), 5275–5303,doi:10.5194/acp-15-5275-2015,2015.

Jacob, D. J., Crawford, J. H., Maring, H., Clarke, A. D., Dibb, J. E., Emmons, L. K., Ferrare, R. A.,Hostetler, C. A., Russell, P. B., Singh, H. B., Thompson, A. M., Shaw, G. E., McCauley, E.,Pederson,J.R.,andFisher,J.A.:TheArcticResearchoftheCompositionoftheTropospherefrom Aircraft and Satellites (ARCTAS) mission: design, execution, and first results, Atmos.Chem.Phys.,10,5191-5212,doi:10.5194/acp-10-5191-2010,2010.

Katragkou,E.,Zanis,P.,Tsikerdekis,A.,Kapsomenakis,J.,Melas,D.,Eskes,H.,Flemming,J.,Huijnen,V., Inness,A.,Schultz,M.G.,Stein,O.,andZerefos,C.S.:Evaluationofnear-surfaceozoneoverEuropefromtheMACCreanalysis,Geosci.ModelDev.,8,2299-2314,doi:10.5194/gmd-8-2299-2015,2015.

Lefever, K., van der A, R., Baier, F., Christophe, Y., Errera, Q., Eskes, H., Flemming, J., Inness, A.,Jones,L.,Lambert,J.-C.,Langerock,B.,Schultz,M.G.,Stein,O.,Wagner,A.,andChabrillat,S.: Copernicus stratosphericozone service, 2009–2012: validation, system intercomparisonandrolesofinputdatasets,Atmos.Chem.Phys.,15,2269-2293,doi:10.5194/acp-15-2269-2015,2015.

Page 71: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page71of148

Parrish, D.F., Derber, J.C.: The National-Meteorological-Centers Spectral Statistical-InterpolationAnalysisSystem,MonthlyWeatherReview,120(8):1747-1763,1992.

Stein,O.,Schultz,M.G.,Bouarar,I.,Clark,H.,Huijnen,V.,Gaudel,A.,George,M.,andClerbaux,C.:OnthewintertimelowbiasofNorthernHemispherecarbonmonoxidefoundinglobalmodelsimulations,Atmos.Chem.Phys.,14,9295-9316,doi:10.5194/acp-14-9295-2014,2014.

Stein, O. and J. Flemming, Documentation of C-IFS-MOZ,MACC-III Deliverable D_24.4,May 2015.Available at http://www.gmes-atmosphere.eu/documents/macciii/deliverables/grg/MACCIII_GRG_DEL_D_24.4_C-IFS_MOZ_final.pdf.

vanderA,R. J.,Allaart,M.A.F., andEskes,H. J.:Extendedand refinedmulti sensor reanalysisoftotalozonefortheperiod1970–2012,Atmos.Meas.Tech.,8,3021-3035,doi:10.5194/amt-8-3021-2015,2015.

Wagner,A.,Blechschmidt,A.-M.,Bouarar, I.,Brunke,E.-G.,Clerbaux,C.,Cupeiro,M.,Cristofanelli,P., Eskes, H., Flemming, J., Flentje, H., George, M., Gilge, S., Hilboll, A., Inness, A.,Kapsomenakis, J., Richter, A., Ries, L., Spangl, W., Stein, O., Weller, R., and Zerefos, C.:Evaluation of theMACC operational forecast system – potential and challenges of globalnear-real-timemodellingwith respect to reactive gases in the troposphere,Atmos. Chem.Phys.Discuss.,15,6277-6335,doi:10.5194/acpd-15-6277-2015,2015.

Page 72: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page72of148

Page 73: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page73of148

7.GlobalAerosols(AER)

WorkintheAERglobalaerosolsubprojectinMACC-IIIhasbeenasmoothcontinuationofthetasksperformedinMACC-II.Ithasproceededalongfourdifferentstrands:

• maintenance,furtherdevelopmentandtestingoftheIFSaerosolschemewithemphasisonthenewIFS-GLOMAPmodalscheme;

• maintenanceandfurtherdevelopmentofthedataassimilationsystem;• furtherdevelopmentandprovisionofsatelliteaerosolretrievals;• further development and provision of global aerosol services (aerosol forcing, aerosol alert

system,aerosolsourceinversion).

Thesestrandsaredescribedinthefollowing.

7.1Maintenance,furtherdevelopmentandtestingoftheIFSaerosolscheme

TheoriginalC-IFS-AERaerosolschemehasbeenmonitoredandmaintainedthroughouttheprojectbuthasseenfairlylittledevelopment.Thecauseofthewell-documentedoverestimationofsulphateaerosols in runs using data assimilation has been investigated using diagnostics on budgets andincrements. The results of this investigation have shown that the problem lies in an imbalancebetweentheinitialstate,providedbythedataassimilationsystem,andtheforwardaerosolmodel.The problem pre-existed the introduction of DMS sources of the sulphate aerosol precursor.Potentialsolutionstoaddressthisproblemwillbetestedinthefuture.

Simulationswithincreasedresolutionhavebeencarriedouttoassesstheimpactofhorizontalandvertical resolution changes on the forecast of aerosols in a biomass-burning event. Thiswork hasbeencarriedoutincollaborationwithNASA,andshowedthatforthisparticularsituation,ahighervertical resolution decreased the optical depth of black carbon whereas a higher horizontalresolutionincreasedit.

Preliminarystepstowardsimprovingtherepresentationofsecondaryorganicaerosols inC-IFS-AERhave been carried out; a review of the state of the art was followed by the acquisition of twodifferentmethodologies,whichwillbeimplementedinthefuture.

Work has been carried out towards the implementation of the nitrate-extendedGLOMAP systemintoC-IFS. Firstly,theplantoimplementacoupledaerosol-chemistryversionofIFS-GLOMAPinC-IFShasbeensuccessfullycarriedout,withthesulphateproductioninGLOMAPbeingderivedfromthe TM5 tropospheric chemistry module in C-IFS. Surface particulate sulphate concentrationssimulated in “coupled aerosol-chemistry C-IFS-GLOMAP” have been evaluated against surfacemeasurements from monitoring stations in Europe, North America and the North Atlantic,comparing also to IFS-GLOMAP runs using the existing fixed-timescale SO2 oxidation approach.Accuratelycharacterisingtheseasonalcycleinsulphateisanessentialfirststeptowardsaccuratelysimulatingnitrateaerosolvariationsbecauseoftherequirementofexcessammoniaabovethatusedup in formation of ammonium sulphate. Initial C-IFS-GLOMAP test runs have confirmed thesuccessfulimplementationoftherequiredextraaerosoltracersforanitrate-extendedGLOMAP.TheadditionalFORTRANsubroutinesrequiredfornitrate-extendedGLOMAPhavealsobeenaddedtoaperforcebranchintheIFSandcompiledsuccessfully.

Page 74: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page74of148

7.2Maintenanceandfurtherdevelopmentofthedataassimilationsystem

Work on updating and testing the aerosol assimilation system has been carried out. Particularattention has been put into upgrading the system to the latest model cycle, and developing thesystemtoincludemoresatellitedatastreamsbothfromactiveandpassivesensors.Workhasalsostartedtocreateanewbackgrounderrorcovariancematrixtoallowtoruntheaerosolsystematahigherverticalandhorizontalresolution.

Progress towards the development of the data assimilation aspects for the IFS-GLOMAP aerosolmodelhasbeenslowerthananticipated.TheobservationoperatorforaerosolopticaldepthintheGLOMAP configuration has been developed. The formulation of IFS-GLOMAP requires specificroutines for the calculation of the aerosol optical depth (observation operator), hence existingroutines couldnot simplybeadapted.Additional fieldsused in inputby theobservationoperatorhadtobeallocatedinthedataassimilation.Thetangentlinearandtheadjointoftheobservationaloperator have been coded but thorough testing is still required. For the time being, a test IFS-GLOMAP configuration has been used in analysis mode without observations, to check that alltechnicaldevelopmentsandscriptchangeswereinplacetoruntheanalysis.Theimplementationinanear-real-timesuiteisstillrequired.

AtestreanalysiswasperformedforthemonthofFebruary2014usingAerosolOpticalDepth(AOD)data from the SEVIRI instrument, processed in near-real-time at ICARE (http://www.icare.univ-lille1.fr/drupal/)usingthealgorithmofThieuleuxetal.(2005).TheinterestintheSEVIRIdatafortheMACC-III project resides mainly in the great coverage provided by the Meteosat geostationarysatelliteview.Thefeasibilityofincludingthisnewdata-streamwasdemonstratedandgoodresultswereobtainedwhenincludingSEVIRIintheassimilation.Moreinvestigationandtheestablishmentofamonitoringsuiteforthisdatasetareneededinordertoexploitthisdatasetinthefuture.

The skill of ECMWF assimilations to predict global distributions of aerosol column amount wasevaluated for the year 2008. Reference data are ground-basedAERONET sun-photometer data ofandmonthlymapsof theMAC climatology.Data assimilation improves theAODdistribution. Thisimprovement,however,dependsstronglytheaccuracyoftheassimilateddata.ThecommonlyusedMODISAODhasastrongpositivebiasatlowerAODsothatatlowAODcasesthepotentialalternateuse of sparser ATSR AOD is competitive. Also investigated are GLOMAP simulations.With strongregionalbiasestocoarsedustandtonon-absorbingaerosolover industrialareas,themodel isnotyetinshapetoreplacethecurrentforecastmodel.

Anew“lidaremulator”forGLOMAPhasbeendevelopedtoderiveattenuatedbackscatterprofilesfrom the model, thereby retaining the full information on size-resolved aerosol composition assimulatedinthemodel.TheapproachwillbetestedinC-IFS-GLOMAPinthefuture.

MACC-III participated to the International Cooperative for Aerosol Prediction (ICAP), a grassrootscommunityofaerosolmodelersanddataproviders,whichprovideaforumtodiscussbestpracticesandfindoptimalcommonsolutionstothechallengesofoperationalglobalaerosolprediction.ICAPhosted sixmeetingswith themes ranging from aerosol observability to aerosolmodel verificationandaerosolensembledataassimilationandprediction.Someofthemeetingshadastrongimpacton the development of new aerosol products to offer to the user community, for example theestablishment of the ICAP multi-model ensemble (Sessions et al. 2014). Likewise, data providershave been engaging activelywithin the ICAP framework to provide enhanced in-situ and ground-basedobservationsofhighqualityandtimelinesstomeettheassimilationandverificationneedsofthis community. ECMWF/MACCwas one of the funding centres of ICAP together with NASA and

Page 75: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page75of148

NRL-Monterey,whichhavestrongaerosolpredictionactivities.NumerousbenefitshavederivedtoECMWF/MACC from participating in the ICAP meetings and discussions, such as increasedinternationalvisibilityoftheatmosphericcompositionactivitiesatECMWF,constantdialoguewithdata providers and possibility to request aerosol products tailored to the MACC needs, andenhanced services to worldwide users through the participation in the ICAPmulti-model, amongothers.

7.3Furtherdevelopmentandprovisionofsatelliteaerosolproducts

Satellitealgorithmsforaerosolretrievalhavecontinuedtobeimprovedanddeveloped.

7.3.1SYNAER

The transfer of the SYNAER methodology from ENVISAT sensors SCIAMACHY and AATSR to theequivalent sensors GOME-2 and AVHRR onboard the METOP platform turned out to be morechallenging than foreseen and could not be achieved inMACC-II. At the start of MACC-III majorsoftware bugs had to be identified and corrected. In the remaining short period ofMACC-III thethree different surface parameterizations were now tested (by radiative transfer calculationsduplicating the retrieval core)andvalidatedagainstAERONETdatawitha smalldataamount.Theradiative transfer calculations show, that the third method (“B factor) has lowest sensitivity todifferent surface types and exhibits a clear sensitivity to increasing AOD. However, the limitedvalidation is not satisfactory. Two possible reasons need further investigation: the iterativecorrectionforaerosol impactonthesurfaceparameterizations,thetreatmentoftheanisotropyofsurfacereflectance.Asaconsequence,itdidnotmakesensetostartNRTdataprocessing.Howeverthe technicalNRTprocessing chain is ready to start andwill be switchedononce the algorithmicissueshavebeensolved.

7.3.2AATSR

After the loss of ENVISAT in April 2012 the provision of NRT AATSR AOD products could not becontinued.InsteadthefocushasbeenontheimprovementoftheAATSRproductand,inparticular,onthetransferoftheAATSRdualviewalgorithmtoapplicationwithSLSTRonSentinel-3(expectedlaunch in the autumn of 2015), to continue a NRT service after the SLSTR commissioning phase.SLSTRandAATSRaresimilarinstrumentbutwithsomedistinctdifferencessuchasthechangeoftheforwardviewonAATSRtoaftviewonSLSTR,whichhasadirect influenceonthescatteringphasefunction.ConsequencesfortheretrievalhavebeenresearchbysimulationsusingMISRdata.MISRprovides both forward and aft viewing angles, but has onlywavebands in theVIS/NIRpart of theelectromagneticspectrum,andhencethefullcapabilityoftheADVapplicationtoSLSTRcouldnotbeinvestigated.

7.3.3SEVIRI

We further evaluated both SEVIRI aerosol products and investigated several leads toimproveproductquality.Overocean,weidentifiedseveralmodificationsthatcanimprovethequalityoftheretrieval.Overland,wemadeprogressonevaluatingtheproductandidentifiedpossibleevolutionsof the science code. Finally, anew algorithm (AERUS-GEO) was explored and implemented forcomparisonandevaluationtootherproducts.

Page 76: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page76of148

The MSG/SEVIRI aerosol product over the ocean and over land has continuedto be generatedroutinelyattheICAREdataandServicesCenter.BothproductsaregeneratedinnearrealtimeatthenativeSEVIRIspatialandtemporalresolution(3kmatnadirand15min).TheyarepubliclyavailablefromICAREFTPsiteandwebsiteinHDFandBUFRformats.

7.3.4IASI

ANRTchainhasbeendevelopedtocalculatedaily(atday-2)10µm(mostlycoarse-mode)AODandmeanaltitude fromIASI/METOP-A.DustAODandheighthavebeenprovidedover landandoceanfor the60°S-60°N-180°W-180°E region, as adaily averageovera0.5°x0.5° grid from June2007 today-2. The dataset is constituted by 1 netcdf file containing daily AOD and mean altitude permonth/year.TheentiredatabaseisreachableviatheARAwebsite.Inparallel,validationisstillon-going,usingAERONETground-baseddatafortheAODandCALIOPdataforthemeanaltitude.

7.3.5Intercomparison

The accuracies of different satellite retrievals for the aerosol column amount in the Earth’satmosphere were investigated. The investigated property is themid-visible aerosol optical depth(AOD).TheinvestigateddatasetsinvolveNASAsensors(MODIS,MISR,SeaWIFS),ESAsensors(ATSR,MERIS)andaCNESsensor(POLDER).Foraselectedtestyear2008,ninedifferentAODretrievalsarecomparedtotrustedreferencedatabyground-basedsun-photometry.Theretrievalskillsaretestedforbias,temporalvariabilityandspatialvariability.Althoughdirectcomparisonsarecomplicatedbysamplingdifferences,anaccuracyrankingisattempted:MISRtakesthetopspot.ThenewimprovedMODIS takes second. ATSR has improved but has not reachedMISR andMODIS. A SeaWiFS likeapproachshouldimproveMERISretrievals.

7.4Globalaerosolservices

7.4.1Aerosolalertsystem

The4-dayMACCforecastofaerosolfieldsfromtheIFSmodelwasusedtoprovideanaerosolalertsystem (with an emphasis on plumes from vegetation fires, dust episodes and pollution build-upsituations). An automated alert classification systemwas developed and tested.Maps of aerosolalert are now generated every day and displayed on a specific MACC/AeroCom website. Aclimatology of aerosol optical depth from the ECMWF-MACC reanalysis is used to identify andclassifyasalertdailyexceedances.

TheprototypeaerosolalertsystemhasbeenavailablesinceApril2014fromtheAeroComwebsite.Alertmapsareavailableback to January2013alongwith infoonaerosolcontributions fromdust,organic aerosol and sulphate. Maps as shown below highlight the aerosol alert regions,corresponding to three alert levels. The alert level 1, 2 and 3 represents aerosol areaswith AODtwice,threeandfivetimestheclimatologicalmean,respectively,andwithAODatleastgreaterthanavalueof0.5.

Aqualitativeevaluationwasperformedagainstobservationsforfourcategoriesofaerosolsevents:smallandlargescaledustevents,fireplumes,andpollutionepisodes.Aerosolevents,whichcametotheattention to theauthorsandhavebeendiscussed in themediaover the last18monthswerechosen. Large dust events andwild fire plumes canmost easily be identified as being above thechosenalert levels.Aerosolalerts relating tosucheventscanbeconsideredas robust.Small scale

Page 77: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page77of148

dusteventsaremore likelytobeoverlookedduetothemesoscalenatureofthephenomena.Thedetectionofalertswithrespecttopollutionepisodesseemstobedifficult.Itwouldrequirefurthereffortstounderstand,whethersuchepisodesaremissedduetomissingsatellitedetection,orduetothechoiceoftheaerosolalertalgorithmchosen.

Contigencytablesareusedtoproposeaquantitativescorefortheperformanceofthealertsystem.AeronetandIFSmodeldatafrom2014areusedtoestablishaerosolalertsatobservingsites,whicharecomparedtothecorrespondingmodelderivedalertsatthesesites.Thetablesrevealthat39%ofthe observed alerts are correctly simulated (ratio true/observed alarms). Observed alerts can befoundonca4%ofalldayswithobservations.Theratio increasedfrom2013to2014from26%to39%.

7.4.2Aerosolradiativeforcingproducts

TheMACC aerosol re-analysis is used to estimate the radiative forcing due to the scattering andabsorptionofsolarradiationbyaerosolsandtothemodificationofthemicrophysicalpropertiesofcloudsbyaerosols.ThetotaluncertaintyassociatedwitheachofthesetwoforcingswascalculatedusingaMonte-Carloapproach,whichaccountsforinteractionsbetweenuncertainparameters.Themethod also allows the quantification of the contribution of different parameters to totaluncertaintyforbothaerosol-radiationandaerosol-cloudradiativeforcing.

The uncertainty on the fraction of aerosols that can be considered anthropogenic is the greatestcontributortototaluncertaintyassociatedwithaerosolradiationinteractions.Uncertaintiesontheresponseofcloudstoaerosolchangescontributemosttototaluncertaintyassociatedwithaerosol-cloudinteractions.Furthermore,althoughaerosol-radiationinteractionscontributethemosttothemeantotal radiative forcing,aerosol-cloud interactionscontributemost to theoveralluncertainty.Changingthereferenceyearusedtocomputetheanthropogenicfractionofaerosolsfrom1750to1850doesnothavealargeimpactonradiativeforcingestimatesattheglobalscale,buthasstrongerregionalimpacts.

7.4.3Aerosolsourceinversion

DustemissionfluxesovertheSaharadesertwereestimatedfromaregionalinversionsystemwhichcombines aerosol optical depth (AOD) retrievals from satellite-borne instruments and aerosolmodelingforaone-yearperiod.WeusedtheatmosphericmodelLMDzcoupledtoanaerosolmodelwith a regional focus on the Sahara desert. The previously existing inversion system has beensignificantlymodifiedandimprovedinordertotakeadvantageofthezoomcapabilityoftheLMDzmodel and its current developments in boundary layer, convective transport and convectivescavengingparameterizations(knownastheNewPhysicspackage).Thedustemissionmodel fromCHIMERE-DUST was implemented and tested in the coupled aerosol and atmospheric model,improving the spatial distribution of AOD over the Sahara desert and downwind regions. WeassimilatedaerosolopticaldepthretrievalsfromthedailycombinedproductofMODISCollection6andestimateddustemissionfluxesovereighteensub-regionsinamonthlybasis.Theseregionsweredefinedbycombiningaclusteringanalysisof themodelemission fluxesandaprioriknowledgeofthemajordustsourceregionsovertheSaharadesert.

Thecostfunctionvalueisreducedintheassimilation,reachingtheexpectedvaluesestimatedbythecost function calculatedwith the linearapproximationof themodel.Additionally, theassimilationsystemiscapabletoreducetheerrorinthesimulatedAODwhenitiscomparedwithAODfromtheAERONETnetwork, and also to reduce theAODbias inmonthswith higher aerosol loadover the

Page 78: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page78of148

regionofstudy.Someaspectsoftheassimilationprocesshavetobeimproved.Inthefirstplace,itisnecessarytoimprovetheestimatesoftheerrorcovariancesBandR,aswellastheassumptionofBas a diagonal matrix. To this effect, the diagnostic proposed by Desroziers et al. have beenimplemented, and correction values forB andRmatrix have tobe estimated. Thenext step is toimprove thedefinitions of B andRmatrices. Afterwards, each emission factorwill be split in twoemission factors: one for fine mode dust and one for coarse dust and super-coarse dust modeslumped together. With this change in the control vector we expect to have an even betterperformanceoftheanalysisAODoverthenorthernSaharaandovertheAtlanticOcean.

The uncertainty on the fraction of aerosols that can be considered anthropogenic is the greatestcontributortototaluncertaintyassociatedwithaerosolradiationinteractions.Uncertaintiesontheresponseofcloudstoaerosolchangescontributemosttototaluncertaintyassociatedwithaerosol-cloudinteractions.Furthermore,althoughaerosol-radiationinteractionscontributethemosttothemeantotal radiative forcing,aerosol-cloud interactionscontributemost to theoveralluncertainty.Changingthereferenceyearusedtocomputetheanthropogenicfractionofaerosolsfrom1750to1850doesnothavealargeimpactonradiativeforcingestimatesattheglobalscale,buthasstrongerregionalimpacts.

Aerosol direct radiative forcing products obtained from the near-real time MACC analysis areavailablefortheperiodAug2012—May2015ontheMACCFTPserverindirectory.

Page 79: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page79of148

8.Globalintegrateddataassimilation,productionandservices(GDA)

TheGlobalDataAssimilation (GDA) sub-projecthasbeencentral to theglobal services inMACC-IIandMACC-III. Ithas strong interactionwith theAER,GHG,GRG,OBS,EMI, FIR, INT,andVAL sub-projectstoensureprovision,qualitycontrol,andfurtherdevelopmentofthevariousglobalservices.Usingitsglobaldataassimilationsystem,GDAprovidedthefollowingglobalservices:

• near-real-timeglobalmonitoringandforecasting• delayed-modemonitoring• reanalyses• user-targetedforecasts

GDAalsoensuredeasyaccess tobothgraphicalanddataproducts through theMACCwebsiteaswellasother formsofdataportals.Emphasis layoncontinuing theprovisionof theservices fromMACC, transforming them to fully operational status, and improvingor newly developing servicesbasedonuserinput.ThetasksofGDAcanbedividedintothreecategoriesandaresummarizedasfollows:

• Integrationofnewdevelopments:- Integrate model and data assimilation developments from other sub-projects in the

global production system as well as a full integration of the MACC system withmeteorologicaldevelopmentsatECMWF

- Assessmentandrefinementoftheintegratedglobalassimilationofdataonatmosphericcomposition

- InteractwithexternalCopernicus-relatedresearchdevelopments• Globalproduction:

- Routine running of a near-real-time (NRT) system for analysis and prediction ofatmosphericcomposition

- Provision of additional global products for extreme cases such as fires or volcaniceruptions

- Routinerunningofadelayed-modesystemforanalysisofatmosphericcomposition- Continuationofthe2003-2010MACCreanalysisfortheyears2011-2012andproduction

ofnewexperimentalreanalysesusingmodelimprovementsandnewdatastreams• Globalservices:

- MonitoringandroutineverificationoftheglobalMACCproductionstreams- Development and maintenance of the MACC website and its systems for display of

resultsandsupplyofdataproducts- Usersupport

OverthedurationofMACC-IIandMACC-IIIGDAhasensuredtheroutinerunningoftheglobaldataassimilationandforecastservices,improvedthesystembasedondevelopmentsfromtheothersub-projects,andmadetheoutputdataavailable toanextensivesetof internalandexternalusers. Inparallel,allrelevantproceduresanddocumentationnecessaryforafulltransitiontotheoperationalphasehavebeenputinplace.

Page 80: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page80of148

Figure8.1.ExampleofglobalMACC-IIIforecastwithCO2(topleft),NO+(topright),CO(bottomleft),andAOD(bottomright).

8.1Integrationofnewdevelopments

AnimportantpartofGDAhasbeentheintegrationandtestingofnewdataassimilationandmodeldevelopments in the ECMWF IFS system. Most modelling components, the chemical transportmodels(CTMs)Mozart,Mocage,andTM5,theaerosolmodelGLOMAP,andthelandsurfacecarbonmodel CTESSEL, have been developed by partners or even other projects. Implementing thesemodelsintheIFSsystemisasignificanttask,asisthesubsequentfine-tuningbasedontestingofthenew model components. On top of this comes the testing of the MACC system when newmeteorological developments are implemented in the IFS system. In order to test the systemdevelopments new tools were developed to enable quick comparison with independentobservations. In addition,MACC-II developed a strong validation component of the global systemwithin the VAL sub-project, which was continued in MACC-III and is described elsewhere. TheValidation reportshavebeen instrumental inassessing thesystemdevelopmentand readiness forimplementationinthepre-operationalforecastingsystem.

DuringMACC-IIasignificantnumberofimprovementshavebeenimplementedaslistedbelow:

• UpgradeofthecoupledMozartCTMtoversionMOZART3.5• Implementationofnewglobalanthropogenicemissiondataset,MACCity• ImplementationofGlobalFireAssimilationSystem,GFAS• Implementation and testing of Composition-IFS (C-IFS) , which allows different chemical

schemestobeused:C-IFS-TM5,C-IFS-Mozart,andC-IFS-Mocage• TestingofC-IFS-TM5indataassimilationmodetopreparefor implementationasnewpre-

operationalsystem• TestingofC-IFS-Mocageindataassimilationmode• CouplingofCO2modellingwithCTESSELlandsurfacecarbonmodelenablinghigh-resolution

globalCO2forecasts

Page 81: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page81of148

• Improvementstoaerosolbinmodel• SupportfordevelopmentandimplementationofGLOMAPmodalaerosolmodel• MigrationofMACC-IIsystemtoIBMandCraysupercomputers• IntroductionofEDA-basedbackgroundstatisticsforgreenhousegasdataassimilation• IntroductionofMOPITTCOobservations• IntroductionofOMINO2andSO2observations• IntroductionofGOME-2O3observationsfromMetop-AandMetop-B• IntroductionofIASICOfromMetop-B• IntroductionofSBUV/2O3observationson21layers• IntroductionofGOME-2NO2,SO2andHCHOobservationsinmonitoringmode• IntroductionofMOPITT,IASIandGOME-2profileobservationsusingaveragingkernels

All these improvements have been implemented in a set of IFS model cycles, in line with theimplementation of meteorological updates to the IFS. The following cycles have been producedduringtheMACC-IIprojectperiod:CY37R3,CY38R1,CY38R2,CY40R1,andCY40R2.

Inaddition,thefollowingimprovementswereintroducedduringMACC-III:

• Implementation of new UV processor reducing the observed systematic errors in thespectrallyresolvedUVproduct

• FurthertestingandimprovementsofC-IFS-TM5,C-IFS-MOZART,andC-IFS-MOCAGE• SupportfordevelopmentandimplementationofGLOMAPmodalaerosolmodel• Implementationoffluxbiascorrection(BFAS)forCO2modelling• IntroductionofIASICOversion6• IntroductionofPMAPAODproduct• IntroductionofGOME-2SO2withvolcanicSO2flag• IntroductionofMODISDeepBlue• IntroductionofassimilationofOCO-2CO2data• ImplementationofnewGFASFRPandemissionarchivingonMARSandintroductionofGFAS

dataserver

ThefollowingcyclehasbeenproducedduringtheMACC-IIIprojectperiod:CY41R1.

Cycle CY40R2, which was developed during MACC-II and introduced as the pre-operationalforecasting system in MACC-III has been a major accomplishment integrating important MACC-IIdevelopments.Especially,theintroductionofC-IFSnowallowsamuchmoreefficientrunningofthechemical forecasting system as well as consistent interaction with the aerosol model, thegreenhousegases,andnot leastthemeteorology intheformofverticaldiffusion,convection,andthe various deposition processes. MACC provides now the first global integrated atmosphericcompositionforecastingsystembasedonaNumericalWeatherPrediction(NWP)model.

8.2Globalproduction

GDAhasalsobeenresponsibleforrunningthevariouspre-operationalglobalsystems.MACCrunsadailydataassimilationand5-dayforecastingsystem,whichassimilatesmostsatelliteobservationsofatmospheric composition that are available in near-real-time as well as all the meteorological

Page 82: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page82of148

observations that are used in the ECMWF NWP system. A significant user group of these dailyforecastsaretheregionalmodellinggroups,bothwithinMACCandoutsidetheproject.Whilemostof the focushasbeenon theEuropeandomain, interest is increasing inAsiaand theAmericasaswell.DirectlinkswithChinahaveforinstancebeenestablishedthroughtheEuropeanUnionfundedPANDA project. The global forecasts are also used for other purposes, most notably the aerosolforecasts,whichare importantforthesolarpower industryandfor(governmental)groupsdealingwithhealthandvisibilityinthemostaffectedcountries.DuringMACC-IIandMACC-IIItheprovisionofthedailyforecastshasbeenverystablewithon-timedisseminationoftheoutputformorethan95%ofthetime.ThistimelinessisbeingmonitoredcontinuouslyandanexampleisshowninFigure.

Figure8.2.Timelinessstatisticsforthenear-real-timeglobalproduction.Thetargettimeisreflectedbythedashedblacklinewhilethetimingoftheforecastrunisindarkgreenandthetimingofthedataarchivingisinlightgreen.

Asecondmajorcomponenthasbeenthereanalysis.MACC-IIhasextendedtheMACCreanalysistoinclude 2011 and 2012. The reanalysis has become a well-recognized data set, especially in thescientific community. It provides an important data set to awealth of users dealingwith climatescience, long-range pollution transport, solar energy applications, and the carbon cycle. Thereanalysis as well as two separate short experimental reanalyses have also supported datareprocessing efforts, such as the ESA CCI initiative. DuringMACC-III a new interim reanalysiswascreated for the period 2003-2014. Aim was to produce a consistent data set for the mentionedperiod,which isbasedonC-IFSandthatcanbeextendedforfutureyears. Inordertosignificantlyspeeduptheproductionprocess,individualyearswereruninparallelwitha1-monthoverlapperiodat each end. This reanalysis will not be used as an officialMACC-III product but can be used byexternalusersonrequest.

A third component has been the support of scientific field campaigns by providing dedicatedforecastsoverspecificdomains.Theseforecastshelpwithflightplanningduringthecampaignandwithinterpretationofthedataafterwards.ThesupportMACChasgiventothesecampaignshasalsohadpositivefeedbackfortheprojectitself.Observationsfromthevariouscampaignsareextremelyusefultofurtherimprovethemodellingsystems.

MACC-II and MACC-III have used a so-called Delayed mode assimilation system to processgreenhousegasobservationsthatarenotavailableinnear-real-time.Originallytheideawastouse

Page 83: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page83of148

these delayed-mode analyses in off-line flux inversions to estimate global CO2 and CH4 fluxes.However,variousissueswiththeobservationsandsomediscrepanciesbetweentheIFSandtheoff-linemodels caused results thatwere not accurate enough to be useful forMACC users. The fluxinversionsarethereforenowusingtheobservationsdirectlyandthedelayed-modesystemhasbeentransformedinatestbedforfuturedevelopments.OneelementofthelatteristheassimilationofGOSATCO2andCH4observations,whichhasmaturedtoapre-operationalstateattheendofMACC-IIIwithaccurateresults(Figure8.3).Anotherelementisthetestingofensembledataassimilationtoestimate assimilation background errors for the greenhouse gas data assimilation. This will alsobenefittheassimilationofreactivegasesandaerosolonthelongerterm.Inparallel,NRTforecastsofCO2andCH4withoutdataassimilationarenowroutinelyprovidedathighresolution.InMACC-IIIboththeforecastanddataassimilationsystemwereimprovedby implementingtheBFASfluxbiascorrection method, developed in GHG, and the timeliness of the delayed-mode system wassignificantlyimproved,nowrunningabout6daysbehindreal-time.

Figure8.3.TimeseriesofXCO2(inppm)overtheTCCONstationsofLauder,LamontandSodankylä.Blackdots:TCCONmeasurements.Cyanline:XCO2fromtheanalysis.Cyandots:XCO2fromtheanalysisusingtheTCCONaprioriinformation.Redline:XCO2fromtworeanalyses.Reddots:XCO2fromthetworeanalysesusingtheTCCONaprioriinformation.

DuringMACC-IIfocuswasalsoputonimprovingtheforecastincaseoflargewildfiresandvolcaniceruptions.Forinstance,thefiresaffectingSingaporeinJune2013andEuropeinJuly2013werevery

Page 84: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page84of148

wellcaptured.Emissionsfromvolcaniceruptionsarenowtosomeextentpickedupautomaticallybythe NRT data assimilation system and mechanisms to introduce sudden emissions for SO2 andvolcanic ash have been developed. This development was continued duringMACC-III resulting ingoodforecastsofsmokeplumesfromCanadianwildfiresaswellascapturingvolcanicSO2emissionsfromtheBardarbungavolcanoonIsland,whichaffectedEuropeanairquality.

8.3Globalservices

Within the global service part of GDA, three important components can be identified: routinemonitoringofincomingandoutgoingdatastreams,productdisplay,disseminationandwebservices,andusersupport.

Forthemonitoringoftheglobalsystem,changeshavebeenincremental.Theroutinemonitoringofincomingsatellitedatahasbeenimprovedallowingquickdetectionofanyissueswiththedata.Thetimelinessof thenear-real-time forecastingsystem isbeingmonitoredaswell.Muchprogresshasbeen made with the routine verification of the various output data streams with independentobservations. Various examples can be accessed on http://www.copernicus-atmosphere.eu/services/aqac/global_verification/.

During MACC-III a new score card has been developed that is used to assess the differences inperformance between the current forecast system (o-suite) and a new upgraded test system (e-suite)basedontheverificationagainstGAWobservations.AnexampleisshowninFigure8.4forthee-suitebasedonCY41R1.ScoresareplottedforO3andCOintheformofmodifiednormalizedmeanbias (MNMB), fractional gross error (FGE) and correlation (CORR).Greenmeans apositive changecomparedtotheo-suite.

Figure8.4.ScorecardforthecomparisonoftheCY41R1e-suiterelativetotheCY40R2o-suite.

Page 85: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page85of148

While theexistingMACCwebsitewasmaintainedwitharoutinegenerationof thousandsofdailyplots, a new structurewas developedwithinMACC-II in collaborationwith the INT sub-project toimprove the user interaction. A major component of the new web site, which went on-line inNovember2012,istheProductCatalogue,whichlistsallavailableMACCproductswitheasyaccesstoplots,data,andvalidation (Figure8.5).DuringMACC-III theunderlyingdatabaseof theProductCatalogue was further developed and now contains INSPIRE compliant metadata. It can alsoautomaticallygenerateproductfactsheetssummarizingthemainmetadataforeachproduct.

Figure8.5.ScreenshotoftheMACCProductCatalogue.

Other aspects of the web site have been redesigned to provide easy-to-understand informationabouttheMACCservicesaswellasdetailedinformationforourroutineusers.Specificsectionshavebeendedicatedtooperationalusersandgeneralusersupport.Thesehavebecomean increasinglyimportantpartofthewebsiteprovidingdetailsaboutsystemchanges,systemstatus,andimpactofimplementedchanges.Webservices facilitating interactionbetweenMACCpartnersalso still existandarebeingexpandedonthe“Internal”partsofthewebsite.

Page 86: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page86of148

Usersupporthasbeenfurtherimprovedbyimplementinganon-linesystem,whichdocumentsuserquestionsandtheresponsestothese.Thisensuresallquestionsarebeingfollowed-upbyrelevantstaff.

8.4Towardsoperations

With the operational phase to be provided by the Copernicus Atmosphere Monitoring Service(CAMS)gettingnearer,ithasbeenanimportantaspectofMACC-IIandMACC-IIItoputinplacethevarious components required for running an operational global forecasting system 24/7. Incollaborationwiththeotherglobalsub-projectsasystemofexperimental(e-suite)andoperational(o-suite) suites has been set-up, and the procedures for developing, implementing and validatingnewcodeandinputdatahavebeendescribedinadocumentthatwasdistributedwithintheproject.In parallel, the same system of e-suites and o-suites is being implemented within the ForecastdepartmentofECMWFtoprepareforthefulloperationalservicein2015.Thedataacquisitionanddata dissemination has gradually been moved to operationally supported systems at ECMWF.Documentationof thevarious componentshasbeenmadeavailableandwill be further improvedandimportantinformationforusersismoresystematicallydistributed.

Page 87: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page87of148

9.Validationactivities(VAL)

9.1Thevalidationsub-project

The goal of the validation (VAL) sub-project inMACC-III can be summarised in the followingway:Everyservice,forecastandreanalysis,ontheMACCwebsiteshouldbeaccompaniedbyavalidationreportwhichisessentialforusers.Thesevalidationreportsshouldberegularlyupdatedandshoulddocument thequalityof the latestmodel/analysis version. Thevalidation subprojecthasprovidedthe validation reports for the core global services of MACC-III. Apart from these reports, VALprovides a set of verification websites hosted by VAL partners with more detailed comparisonsbetween the MACC service products and individual validation datasets. Furthermore, real-timeverificationstatisticsisproducedinanoperationalway,incollaborationwiththeglobalproductionteam(GDA),andismadeavailableontheMACCwebsite.Quick-lookvalidationworkwasperformedtoprovidefeedbacktothemodellingteamsincaseofupdates:thee-suiteverificationreports.

Validation has to reach an operational status, which implies an automated production ofverification/validation data and plots, and updates of validation reports with fixed intervals. Thefocus of VAL is on the global reactive gas and aerosol services. The regional sub-projects ENS(regional ensemble forecasts) and EVA (regional reanalyses and assessments) as well as thegreenhouse gas sub-project have their own validation activities and reports. The production andpresentationofvalidationresultsaspresented inthereportontheNear-RealTimeglobalforecastsystemisharmonised.

The Validation (VAL) sub-project is maintaining interaction with most of the other MACC sub-projectstoensureroutinedocumentationofthequalityoftheglobalserviceswithcontributionstothe regional air quality sub-projects. The other MACC-III sub-projects have be involved to unifyscoringapproachesandpresentationofthevalidationresults.Inaccuraciesinthefireemissionandother emission input will be quantified based on their effects in the various global and regionalatmosphericanalysesandforecasts.

Below we will provide an overview of the main activities of the validation team. The reportsmentionedinthetextcanallbefoundontheMACCwebsite.

9.1.1Structureofthevalidationreports

Thevalidationreportsarewrittenfortheusers,andprovidearegularlyupdateddocumentationofthe validation results for the service data products available in theMACC catalogue. The reportscontain:

• Asummarywithanoverviewofthemainfindings.Thissummaryissplitaccordingtoareasof interest to users: Climate forcing, regional air quality, ozone layer and UV. SpecificattentionisgiventotheabilityoftheMACCsystemtocapturerecentevents.

• A section with detailed information on the model configuration, analysis system andevolution/upgradedetails.

• Availabilityandtimingstatistics.• A detailed presentation of the validation results. This is the bulk of the reports. A reader

interested in the details of certain aspects of the system can directly jump to thecorrespondingsub-sectionofthereport.

• Anannexwithdetailsonthevalidationmethodology.

Page 88: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page88of148

The reports are delivered on a regular basiswith an update frequency depending on the service:every3monthsforthereal-timeforecasts,every6monthsduringproductionofthereanalysis,andaftermajorupgradesforotherserviceproducts.

Figure9.1.E-suiteverificationexampleoftroposphericozonefortheperiodDecember2013–May2014,withsolidlinesshowingtheo-suitebasedonIFS-MOZARTatthattime,andthedashedlinesthee-suitebasedonC-IFS-CB05.AclearimprovementisseenovertheAntarcticandArcticregions,whilesimilarperformancewasreachedovertheNHmid-latitudesandtropics.Thisexampleistakenfromthee-suitereportofSeptember2014.

9.1.2E-suiteevaluationreports

DuringMACC-IItheprocessofintroducingnewupgradeshasevolvedtoprepareforoperations.Fortheglobalnear-realtimecompositionforecast/monitoringservice,upgradesaretypicallyintroducedafewtimeseachyear.Thestepsare:

• Anewmodelversionand/orassimilationupgradeispreparedandevaluatedoff-line.Fortheglobalcompositionforecastservice,thisworkisperformedbythereactivegas,aerosolandproductionsub-projects(GRG,AERandGDA).

• Thenewsystemversionisdocumented,andaso-called"e-suite"ispreparedandisrunfortypicallyhalfayearinparalleltotheoperational"o-suite".

• Theperformanceofthise-suiteisevaluatedagainsttheo-suitebytheVALteam.Basedonthe validation results a recommendation is given, whether or not to replace the o-suite.Typicallyitisrequiredthatthee-suiteperformsequallyorbetteronallaspectsthantheo-suite.

The VAL team has produced two e-suite evaluation reports during MACC-III. The first reportdiscussed the significantupgradeof theglobalanalysisand forecast system fromthecoupled IFS-MOZARTapproachtowardstheintegratedchemistrybasedonComposition-IFS(C-IFS),usingTM5/CB05 chemistry for the troposphere. In the second report updates largely due to modifiedmeteorology was assessed, as well as specific changes in the data-assimilation system related toaerosol.

Page 89: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page89of148

9.1.3Scoring

Part of the research activity inVAL consists of definingmeaningful validation scores. In particularthere isaneedtodefinescores,whichcandocumentthe improvementsof thesystemover time,andwithrespecttoothermodellingsystems.

TheVALsub-projecthasproducedthereport"SkillscoresandevaluationmethodologyfortheMACCIIIproject".Thisdocumentsummarisesthegeneralscoresandevaluationmethods,whichhavebeenusedduringMACCIII,andisanupdateofthereportproducedinMACCII.Feedbackfromtheteamsworkingonvalidation,modellingteamsandusersareincorporatedinupdatesofthisdocument.

Figure9.2.SnapshotimpressionsfromtheVALverificationpages.

Inthisdocumentspecialattentionwasgivenonthe"headlinescore"topic.Thepurposeofthisnoteistoguidethedevelopmentofasetofsummaryskillscoreswhichcanbeusedtodocumenttheperformance,andmonitortheimprovementsoftheMACC-III/CAMSsystemovertime.TheseskillscoresaretargetingasetofCopernicus-atmosphereuserapplicationareas.

Page 90: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page90of148

9.1.4Validation/verificationwebpages

Asetof12dedicatedwebsiteshasbeensetup,wheremoredetailedcomparisonsagainstindividualdatasetsarepresented. Thesewebsitesaremaintainedby theVALpartners, and canbeaccessedthrough the MACC webpage http://copernicus-atmosphere.eu/verification-global-services. AnimpressionofthedatasetsprovidedisgiveninFigure9.2.Thewebsitescontain:

• Verificationplotsfortheglobalatmosphericmonitoringandforecastingservice.• Verificationplotsforthereanalysisservice.• Real-timemonitoringplots,providedbyECMWF.

During MACC-II the VAL subproject has started a collaboration with the EU NORS project(http://nors.aeronomie.be)tomakereal-timemeasurementsfromtheNDACCnetworkavailabletoMACC. The NORS server, http://nors-server.aeronomie.be, provides a comprehensive browser ofplotscomparingNDACCdatawiththeMACCglobalanalyses/forecasts.

9.1.5Casestudies:fire,dustandpollutionevents,ozonedepletion

OneofthemainaimsoftheMACCsystemistoprovideanalysesandforecastsofextremeevents.Theevaluationof suchevents isa cross-cutting topic,and is co-ordinatedwith theother involvedsub-projects,andinseveralcasesrapid-responseactionshavebeentaken.Examplesaremajorfires(Wild fires in Australia, February 2015), dust storms, volcano eruptions (Bárðarbunga eruption in2014) andairpollutionevents. TheVALgroup studiedmore than10events, and the resultshavebeenincludedinthevalidationreports.AnexampleisshowninFigure9.3.

Figure9.3.Casestudyofawildfireplumewithstrongcarbonmonoxide(CO)emissionsoriginatingfromAustralia,February2015.TheCOplumewasobservedbyIASIsatelliteinstrument(left),andmodeledwiththeMACCo-suite(right).ThetransportoftheCOplumewaswelldescribed,butthetotalamountwasunder-estimatedby50%comparedtotheobservations.

Data from large internationalmeasurement campaigns is alsoused toevaluate themodel results.ExamplesarePOLARCAT/POLMIP,PEGASOS(EU)orSEAC4RS(NASA).

Page 91: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page91of148

9.1.6Validationresearch

Researchhasbeenconductedonthefollowingtopics:

• Analysisofevents,seeabove.• Improvementofthevalidationmethodology• Developmentandharmonisationofscoringapproaches.• Evaluationofnewdatasets:ceilometerandMAX-DOAS.

Thedetailedvalidationproceduresaredescribed in theannexesof thevalidation reports, andwereferthereadertothosereportsformoredetails.

ThevalidationresultsbythepartnersofVALhavebeendescribedinscientificpapers.ThesemaybegroupedindedicatedvalidationpapersthatarecontributionstotheMACCspecialissue(Lefeveretal.,2014;Katragkouetal.,2015;Cuevasetal,2014;Wagneretal.,2014)orcontributionstopapersledbypartnersfromothersub-projectsofMACC(Innessetal.,2015;Flemmingetal.,2015;PérezGarcía-Pando et al., 2014; Stein et al., 2014; Cesnulyte et al., 2014). An overview of the MACCvalidationactivitiesisprovidedinEskesetal.(2015).

DailyECMWFarchiveretrievalsofaerosolprofilesandboundary layerheightshasbeensetup forthe locations of selected DWD ceilometer stations. A scientific paper is in preparation on theevaluationofMACCsimulationsofCanadianForest fireswithceilometerprofiles. (Flentjeetal., inpreparation,2015).

Figure9.4.AnnualaverageofGLOMAPtotalAOD,anditsbiaswithrespecttoAERONETobservations.Redindicatesanover-estimateofGLOMAP.

The VAL team invited the regional modelling teams (from the ENS, EVA, and EDA subproject) toparticipate in a pilot study, comparing the modelling results with the MAX-DOAS measurementsfromseveralsites inEurope.AmajorityoftheregionalairqualityteamscommittedthemselvestocontributeandthecollaborationwasfurtherdiscussedduringtheDecember2014regionalmeeting.UsingMAX-DOASmeasurementsforMACC-IIImodelvalidationispromisinganditisproposedtouseMAX-DOAS in future validation activities of the operational COPERNICUS atmospheric service, inparticularfortheregionalmodels.

Page 92: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page92of148

TheGLOMAPmodeloutputfortheyear2008hasbeenevaluatedintermsofitsopticalproperties,including total aerosol optical depths (AOD), and its fine and coarse mode fraction. Also theabsorptioncomponent(AAOD)isindependentlyassessed.Togetherthisprovidesinformationabouttheaerosolsizeassociatedwiththeaerosolamount.

ItwasfoundthatthetotalAODisabitover-estimatedoverindustrialregions,likelyduetobiasesinthefine-modefraction.Ontheotherhand,theAAODwasunder-estimated.

9.2ValidationoftheMACCnear-realtimeglobalatmosphericcompositionservice

Thenear-realtimeglobalatmosphericcompositionserviceisoneofthecoreservicesofMACC-III.Itis providing daily (near-real time) forecasts and analyses of the global reactive gas and aerosoldistribution.OneofthemaintasksofVAListheevaluationoftheproductsfromthisservice.DuringMACC-III, regular updates of the near-real time reports were produced every threemonths. Thestructureofthesereportswasdiscussedabove.ThelastMACC-IIIreportiscalled"Validationreportof the MACC near-real time global atmospheric composition service: System evolution andperformancestatistics, statusup to1March2015"Thevalidation reportsarepubliclyavailableathttp://copernicus-atmosphere.eu/quarterly_validation_reports.

Thefollowingindependentdatasetshavebeenusedtoproducethevalidationreports:

• ProfilesofCOandO3fromMOZAIC/IAGOS.• SurfaceobservationsofCOandO3from13GAWstationsand2EMEPstations.• OzoneprofilesfromO3sondes.• AssimilationresultsfromtheBASCOE,SACADA,andTM3DAMdataassimilationsystems.• SatellitedatafromACE-FTS,Odin-OSIRISandOMPS-limb.• MOPITTandIASICOobservations.• RetrievalsofNO2andHCHOcolumnsfromtheSCIAMACHYandGOME-2instruments.• Tenground-basedstationdatafromESRLGlobalMonitoringDivisionnetwork

(http://www.esrl.noaa.gov/gmd/).• AerosolopticaldepthandAngströmExponentdatasetsfromtheAeronetsunphotometer

network,availablefromNASAGoddard.SupportinggraphsweregeneratedwiththeAeroComtools.

• AnalysesofdustplumescloudsinNorthernAfricaandEuropehavebeenperformedwithAODfromAERONETstations,withMODIS(Aqua/Terra)andwithlidarprofiles.

• Near-realtimemonitoringoftheglobalforecastsisbasedonNRTobservationsfromtheAERONETnetworkandozoneandcarbonmonoxidefromWMOGAWsurfacestations.

• Ground-basedgreenhousegasobservationsfromtheICOSprojectandTCCONnetwork.• NDACCnetworkreal-timedataprovidedbytheNORSproject.

Page 93: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page93of148

Table9.1:OverviewofthetracegasspeciesandaerosolaspectsintheMACC-IIIvalidationreport.ShownarethedatasetsassimilatedintheMACCNRToperationalsystem(secondcolumn)andthedatasetsusedforNRTvalidation(thirdcolumn).Greencolorsindicatethatsubstantialdataisavailabletoeitherconstrainthespeciesintheanalysis,orsubstantialdataisavailabletoassessthequalityoftheanalysis.Yellowboxesindicatethatmeasurementsareavailable,butthattheimpactontheanalysisisnotverystrongorindirect(secondcolumn),orthatonlycertainaspectsarevalidated(thirdcolumn).

Species,verticalrange

Assimilation Validation

Aerosol,opticalproperties

MODISAqua/TerraAOD AOD,Ångström:AERONET,GAW,Skynet,MISR,OMI,lidar

O3,stratosphere

MLS,GOME-2A,GOME-2B,OMI,SBUV-2

Sonde,lidar,MWR,FTIR,OSIRIS,OMPS,BASCOEandMSRanalyses

O3,UT/LS

Indirectlyconstrainedbylimbandnadirsounders

IAGOS,sonde

O3,freetroposphere

Indirectlyconstrainedbylimbandnadirsounders

IAGOS,sonde

O3,PBL/surface

- Surfaceozone:WMO/GAW,NOAA/ESRL

CO,UT/LS

- IAGOS

CO,freetroposphere

IASI,MOPITT IAGOS,MOPITT,IASI

CO,PBL/surface

IndirectlyconstrainedbysatelliteIRsounders

SurfaceCO:WMO/GAW,NOAA/ESRL

NO2,troposphere

OMI,partiallyconstrainedduetoshortlifetime

SCIAMACHY,GOME-2,UV-VisDOAS

HCHO

- GOME-2,UV-VisDOAS

SO2

OMI(Individualvolcaniceruptionsandstrongsources)

-

Stratosphere,otherthanO3

- NO2columnonly:SCIAMACHY,GOME-2

UV-Index ConstrainedbyassimilationofozoneandaerosolAOD

COSTUVIndexDatabase

Foragoodunderstandingofthevalidationresultsitisimportanttoknowwhichaspectsoftheglobalassimilationsystemareconstrainedbytheobservations,andwhichaspectsarecoveredbythevalidationdatasetsused.ThisissummarisedinTable9.1.

Figure9.5providesexamplesoftheevaluationofaerosolAODandNO2.Formoredetailswerefertothereports.

Page 94: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page94of148

Figure9.5.(left)Normalizedmeanbias(%)ofAerosolOpticalDepthofMACC_osuite(thickredcurve)againstAERONETdata;AlsogivenistheMACC_osuiteatlastforecastday(lightredcurve);MACC_cntrl(bluedashedcurve);MACC_cntrlatlastforecastday(lightbluedashedcurve).(right)TimeseriesofaveragetroposphericNO2columns[1015moleccm-2]fromSCIAMACHY(uptoMarch2012)andGOME-2(fromApril2012onwards)comparedtomodelresultsforEurope.ExamplestakenfromtheMACC-IIINRTvalidationreportedition4ofMay2015.

9.3ValidationoftheMACCglobalreanalysis

The global reanalysis for the years 2003-2012 has been evaluated in validation reports producedduringMACC-II. The last issue of theMACC-II reanalysis report is called "Validation report of theMACCreanalysisofglobalatmosphericcomposition:Period2003-2012"Thevalidation reportsarepubliclyavailableat:

http://www.gmes-atmosphere.eu/services/aqac/global_verification/validation_reports/

WithinMACC-III dedicated activities have been performed to extend the evaluation of theMACCreanalysis in termsofsurfaceozone(Katragkouetal.,2015)andozone in theUTLS (Gaudeletal.,submitted to Tellus, 2015). In both cases, clear improvements due to assimilation are found, asillustratedinFigure9.6.

DuringMACC-IIIanevaluationofthereanalysiswithCARIBICaircraftdatafortheperiod2006-2012has been set up using the central IAGOS data base. In this way an automatic verification overdifferentairportsisperformed,availablefromhttp://www.iagos.fr/maccwebsite.CARIBICprovidesdata over airports not sampled by IAGOS-CORE or at different time periods, allowing for amoreglobalevaluationwithin-situobservations.

WithsupportofESA’saerosolCCI initiativeanew longtermaerosolclimatedatarecordbasedonATSRsensordatawasdeveloped.Thisnewdatarecordforthemid-visibleaerosolopticaldepthAOD(at 550nm) is competitive to commonly applied data records of NASA EOS sensors (e.g. MODIS,MISR)butgoes furtherback in time,until late1995. Thisdata record representsan independentdata source for theevaluationof theMACC reanalysis. TheMACC reanalysis consistently showsapositivebiasovertheocean,while insomelandareasanunderestimateofAODwasfound,whichsuggestedasystematicbiasofeitherATSRortheMACCreanalysisagainstthetrueAOD.

Page 95: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page95of148

Figure9.6.TimeseriesofmonthlymeanozoneintheLowerStratospherefortheMACCReanalysis(red),thecontrolrun(green)andMOZAIC-IAGOSobservations(blue).Thebarsdenotethestandarddeviations(2σ)forMOZAIC-IAGOSobservations.

9.4ValidationofotherMACCservices

9.4.1Theozonemulti-sensor43-yearreanalysis

TheOzoneMulti-Sensorreanalysisversion2(MSR-2)(vanderAetal.,2015)providesglobalozonecolumn field timeseries, for theperiod1970-2012,onagridwith0.5x0.5degree resolutionandwitha time intervalof 6hours. Thisdetaileddata set is producedbyassimilatingall independentsatellitecolumnobservationdatasetspubliclyavailable(15datasetsintotal,BUV-Nimbus4,TOMS-Nimbus7,TOMS-EP,SBUV-7,-9,-11,-14,-16,-17,-18,-19,GOME,SCIAMACHY,OMIandGOME-2).Thesesatelliteretrievalshavebeencalibratedagainsta"groundtruth"consistingofozonecolumnmeasurementsfromthenetworkofBrewerandDobsoninstruments.

In the report with title "Validation report of theMACC 43-year multi-sensor reanalysis of ozonecolumns, version 2, Period 1970-2012", the MSR-2 fields are compared with individual Brewer-Dobson measurements, and with individual satellite observations. Observation-minus-forecaststatisticsisusedtostudytheinternalconsistencyoftheozoneanalyses,thedetailederrorestimatesandthesatellitedata.

9.4.2EvaluationofBASCOEstratosphericanalyses

TheBASCOEassimilationsystem,operatedbyBIRA-IASB,isusedintheMACC-IIIprojecttoassimilatethe offline observations of stratospheric composition by Aura-MLS. The quality of the BASCOEanalyses of O3, H2O, HCL, HNO3, N2O and CLO have been verified by comparison with theassimilated observations and validated through the comparison of a preliminary run withindependentobservationsbyACE-FTSfortheyear2011.Itwasfoundthatforallassimilatedspeciesrelativebiasesarealwaysbelow10%exceptinregionsandseasonswheretheabundanceissmall.Comparison toACE-FTS shows that theBASCOEanalyseshavea verygoodquality. Specifically forozone, biases between analyses and assimilated observations are less than 3% in the lowerstratosphereandlessthan5%atthemiddlestratosphere.

Page 96: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page96of148

Figure9.7.MACCaerosolalertmapfor31July2014,withextendedfireplumesoverNorthAmerica(right).

9.4.3EvaluationofAerosolAlertService

As a combined AER / VAL effort an aerosol alert service has been setup and evaluated, with anemphasis on plumes from vegetation fires, dust episodes and pollution build-up situations. Alertlevels1,2and3aredefined to representaerosol areaswithAOD twice, threeand five times theclimatologicalmean,respectively,andwithAODatleastgreaterthanavalueof0.5.

Aqualitative evaluationwasperformedagainst observations, specifically for aerosol eventswhichhavebeendiscussedinthemedia.Itwasfoundthatlargedusteventsandwildfireplumescanmosteasilybeidentifiedasbeingabovethechosenalertlevels,anditsaerosolalertscanbeconsideredasrobust,seeFigure9.7.Small-scaledusteventswerefoundtobemorefrequentlyoverlookedduetothemeso-scalenatureofthephenomena.Thedetectionofalertswithrespecttopollutionepisodesseems to be difficult. It would require further efforts to understand whether such episodes aremissedduetomissingsatellitedetection,orduetothechoiceoftheaerosolalertalgorithmchosen.

Contingency tables provide a quantitative score for its performance, using Aeronet observationsfrom2014.Thetablesrevealthat39%oftheobservedalertsatthesesitesarecorrectlysimulated(ratiotrue/observedalarms).Observedalertscanbefoundonca4%ofalldayswithobservations.The ratio increased from2013 to 2014 from26% to 39%.All figures are accessible via theMACCaerosolalertwebinterface:

http://aerocom.met.no/cgi-bin/aerocom/surfobs_annualrs.pl?Parameter0=ALERT_AER

Page 97: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page97of148

9.5References

vanderA,R.J., Allaart,M.A.F., and Eskes,H.J.: Extended and refined multi sensor reanalysis oftotalozonefortheperiod1970–2012,Atmos.Meas.Tech.,8,3021-3035,doi:10.5194/amt-8-3021-2015,2015.

Cesnulyte, V., Lindfors, A. V., Pitkänen, M. R. A., Lehtinen, K. E. J., Morcrette, J.-J., and Arola,A.:Comparing ECMWF AOD with AERONET observations at visible and UVwavelengths,Atmos.Chem.Phys.,14,593-608,doi:10.5194/acp-14-593-2014,2014.

Cuevas, E., Camino, C., Benedetti, A., Basart, S., Terradellas, E., Baldasano, J.M.,Morcrette, J. J.,Marticorena,B.,Goloub,P.,Mortier,A.,Berjón,A.,Hernández,Y.,Gil-Ojeda,M.,andSchulz,M.:TheMACC-II2007–2008reanalysis:atmosphericdustevaluationandcharacterizationovernorthernAfricaandtheMiddleEast,Atmos.Chem.Phys.,15,3991-4024,doi:10.5194/acp-15-3991-2015,2015.

Eskes,H.,Huijnen,V.,Arola,A.,Benedictow,A.,Blechschmidt,A.-M.,Botek,E.,Boucher,O.,Bouarar,I., Chabrillat, S., Cuevas, E., Engelen, R., Flentje, H., Gaudel, A., Griesfeller, J., Jones, L.,Kapsomenakis,J.,Katragkou,E.,Kinne,S.,Langerock,B.,Razinger,M.,Richter,A.,Schultz,M.,Schulz, M., Sudarchikova, N., Thouret, V., Vrekoussis, M., Wagner, A., and Zerefos, C.:Validation of reactive gases and aerosols in theMACC global analysis and forecast system,Geosci.ModelDev.Discuss.,8,1117-1169,doi:10.5194/gmdd-8-1117-2015,2015.

Flemming,J.,Huijnen,V.,Arteta,J.,Bechtold,P.,Beljaars,A.,Blechschmidt,A.-M.,Diamantakis,M.,Engelen,R.J.,Gaudel,A.,Inness,A.,Jones,L.,Josse,B.,Katragkou,E.,Marecal,V.,Peuch,V.-H., Richter, A., Schultz,M. G., Stein, O., and Tsikerdekis, A.: Tropospheric chemistry in theIntegratedForecastingSystemofECMWF,Geosci.ModelDev.,8,975-1003,doi:10.5194/gmd-8-975-2015,2015.

Gaudel,Audrey,HannahClark,ValerieThouret,LukeJones,Antje Inness, JohannesFlemming,OlafStein, Vincent Huijnen, Henk Eskes, Philippe Nédélec, Damien Boulanger, On the use ofMOZAIC-IAGOSdatatoassesstheabilityoftheMACCReanalysistoreproducethedistributionofO3andCOintheUTLSoverEurope,submittedtoTellusB,2015.

Huijnen, V., Flemming, J., Kaiser, J.W., Inness, A., Leitao, J., Heil, A., Eskes, H. J., Schultz,M. G.,Benedetti, A., Dufour, G., and Eremenko, M., Hindcast experiments of troposphericcompositionduringthesummer2010firesoverWesternRussia,Atmos.Chem.Phys.12,4341-4364,doi:10.5194/acp-12-4341-2012,2012.

Inness, A., Blechschmidt, A.-M., Bouarar, I., Chabrillat, S., Crepulja, M., Engelen, R. J., Eskes, H.,Flemming, J.,Gaudel,A.,Hendrick,F.,Huijnen,V., Jones,L.,Kapsomenakis, J.,Katragkou,E.,Keppens,A.,Langerock,B.,deMazière,M.,Melas,D.,Parrington,M.,Peuch,V.H.,Razinger,M.,Richter,A.,Schultz,M.G.,Suttie,M.,Thouret,V.,Vrekoussis,M.,Wagner,A.,andZerefos,C.:Dataassimilationofsatellite-retrievedozone,carbonmonoxideandnitrogendioxidewithECMWF's Composition-IFS, Atmos. Chem. Phys., 15, 5275-5303, doi:10.5194/acp-15-5275-2015,2015.

Katragkou,E.,Zanis,P.,Tsikerdekis,A.,Kapsomenakis,J.,Melas,D.,Eskes,H.,Flemming,J.,Huijnen,V., Inness, A., Schultz,M.G., Stein,O., and Zerefos, C. S.: Evaluation of near-surface ozone

Page 98: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page98of148

overEuropefromtheMACCreanalysis,Geosci.ModelDev.,8,2299-2314,doi:10.5194/gmd-8-2299-2015,2015.

Langerock,B.etal.Descriptionofalgorithmsforco-locatingandcomparinggriddedmodeldatawithremote-sensingobservations.Geosci.ModelDev.8,911-921(2015).

Lefever, K., van der A, R., Baier, F., Christophe, Y., Errera, Q., Eskes, H., Flemming, J., Inness, A.,Jones,L.,Lambert,J.-C.,Langerock,B.,Schultz,M.G.,Stein,O.,Wagner,A.,andChabrillat,S.:Copernicus stratospheric ozone service, 2009–2012: validation, system intercomparison androlesof inputdatasets,Atmos.Chem.Phys.,15,2269-2293,doi:10.5194/acp-15-2269-2015,2015.

PérezGarcía-Pando,C.,M.C.Stanton,P.J.Diggle,S.Trzaska,R.L.Miller,J.P.Perlwitz,J.M.Baldasano,E. Cuevas, P. Ceccato, P. Yaka, and M.C. Thomson, 2014, Soil Dust Aerosols and Wind asPredictors of Seasonal Meningitis Incidence in Niger, Environmental Health Perspectivesdoi:10.1289/ehp.1306640.

Stein,O.,Schultz,M.G.,Bouarar,I.,Clark,H.,Huijnen,V.,Gaudel,A.,George,M.,andClerbaux,C.:OnthewintertimelowbiasofNorthernHemispherecarbonmonoxidefoundinglobalmodelsimulations,Atmos.Chem.Phys.,14,9295-9316,doi:10.5194/acp-14-9295-2014,2014.

Wagner, A., Blechschmidt, A.-M., Bouarar, I., Brunke, E.-G., Clerbaux, C., Cupeiro, M.,Cristofanelli,P.,Eskes,H.,Flemming,J.,Flentje,H.,George,M.,Gilge,S.,Hilboll,A.,Inness,A.,Kapsomenakis, J., Richter, A., Ries, L., Spangl, W., Stein, O., Weller, R., and Zerefos, C.:EvaluationoftheMACCoperationalforecastsystem–potentialandchallengesofglobalnear-real-timemodellingwith respect to reactive gases in the troposphere, Atmos. Chem. Phys.Discuss.,15,6277-6335,doi:10.5194/acpd-15-6277-2015,2015.

Page 99: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page99of148

10.DataassimilationforEuropeanairquality(EDA)

The ability to provide analysed chemical fields, along with locally and temporally resolved errorestimates,isanessentialcomponentofCAMSregionalairqualitymulti-modelensembleapproach.Thisconcernspotentiallyall routinelymeasuredchemical speciesandparticles, irrespectiveof theLevel-1 (direct) or Level-2 (retrievals) nature of the observations . The resulting analyses of thechemical state of the atmosphere are required to be chemically consistent and to preserve theintegrityofnaturalbalancewithinchemicalfamilies.

Fourhigh-levelobjectiveshavebeenachievedduringMACC-IIIaspartofEuropeanairqualityDataAssimilationactivities(EDA):

• (1)Moredatafromspacebornesensorswereassimilated;• (2) The prototype operational data assimilation algorithms were improved in terms of

operationalstability,efficiency,robustness,andaccuracy;• (3)Thequalityofthedataassimilationproductswasestimated;• (4)Data assimilationalgorithmswereenabled to account for special orhardlypredictable

events(mineraldustoutbreaks,forestfireemissionsandvolcanicemissionsspecifically).

Buildingonexistingcode,partnersfurtherdevelopedtheirowndataassimilationcodeindividually,whichareofvariationalorcomplexity reducedKalman filter type,bothofBLUE. Inanalogy to themulti-modelensemblesusedinR-EVAandR-ENS(seechapter12and11respectively),thediversityof existing assimilation algorithms at individual centres was maintained, apart from furtherdevelopments,whereexchangeofmodulesandcommonimprovementswereimplemented.

ThefollowingchemicaldataassimilationalgorithmshavebeenusedinMACC-III:

• RIUUK:The3Dor4D-variationaldataassimilationmethodincludingextensionforemissionrate inversion is applied, dependent on objective. With the diffusion approach, asophisticated covariancemodellingmethod is operational, as observation operators for anumberofsatellitesensors,includingCALIOP.

• LISA:TheEnsembleKalmanFilter(CHIMERE-EnKF)usingasquarerootformulation(insteadof directly perturbing observations) is applied for the ozone assimilation of surface andsatellitedata.Also,assimilationofNO2satellitedatahasbeentestedisapplied.

• KNMI/TNO: Ensemble Kalman filterwith specification of uncertainty ofmodel parameters(emissions,boundaryconditions,depositionrate).

• FMI:Variational schemeAQanalyses: currentlyO3,NO2,SO2areassimilatedhourly,PM inpreparation. Emission inversion: identification of pollen sources, season-wide assimilationwindow.

• MF-CNRM/CERFACS:thealgorithmisbasedonthe3D-Variationaldataassimilationmethod.It is configured to optimize the initial condition of chemical species with one hourforecast/analysiscycles. Itallowstheassimilationof in-situmeasurements,verticalprofilesand partial/total columns of gas/aerosols. Full 3D background error covariances aremodelled with a diffusion equation approach and can be specified through ensemblesimulations.

• TheMATCHdataassimilationsystematSMHI isa3Dvariationalschemesolved inspectralspace.ThespectralbackgrounderrorsarederivedbytheNMCapproachforathreemonthtraining period. Observation operators are included for in-situ observations and satellitedatasuchasAIRS,GOME2,IASI,MODIS,MOPITTandOMI.

Page 100: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page100of148

• EMEPatmet.nousesintermittent3DVARdataassimilationofNO2columnsfromOMIandsurface concentrations from in situ NO2 and O3 measurements near-real time. DataassimilationforAODiscodedinEMEPforallPMvariablesthatcontributetoAOD.

WhileEDAprimarilyservesENSandEVA, it isalso linkedwiththeMACC-III subprojects Integratedglobaldataassimilation,productionandservices(GDA),Globalreactivegases(GRG),Globalaerosols(AER),Acquisitionofobservations(OBS),Emissions(EMI),andFiredataassimilation(FIR).

TheorganisationofworkduringMACC-IIIfollowedthefourabove-mentionedobjectives.

10.1Satellitedataassimilation

10.1.1NO2columndataassimilationmodules

NO2 tropospheric columns from SCIAMACHY (meanwhile obsolete), OMI and GOME-2, wereintegrated in the data stream, and the related observation operator was developed or updated,makinguseofaveragingkernel information(seeFigure10.1foranexample). DedicatedOMINO2datasetsfortheperiod2004-2013havebeengeneratedbyKNMItobeusedintheEVAreanalyzesand were made available to the project partners through the MACC-III web site.Foruseofthisproduct intheLOTOS-EUROSassimilation/forecastsystem,anobservationoperatorhasbeenimplementedbasedontheguidelinesprovidedwiththeproduct.ThishasbeenappliedforEURAD-IM(RIUUK)aswell,whileforGOME-2owndevelopmentsweretaken.

Figure10.1.ExampleofOMItroposphericNO2productprovidedtoMACCregionalmodelsystemsbyKNMI(left).Calculatedbias(observations–model)for15thAugust2009withCHIMERE.Thebiasvariesbetween(-10and10)*1015molecules/cm2(right).

Page 101: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page101of148

Carewastakenthaterrorsduetolimitedmodeltopheightdonotdegradetheassimilationresult.SinceNO2andO3arecloselyrelated,CERFACSevaluatedtheimpactoftheanalysesofthosegasesusingthe3D-VAR/Valentinasuite.LISAusesthesameobservationoperatorthathasbeendevelopedforsatelliteozoneobservationstoassimilate OMI observations. For a few days case study, large biases between observed andsimulatedcolumnshavebeenobservedexplainingthatnoimprovementofNO2andO3surfacefieldshavebeenobserved.

10.1.2COdataassimilationmodule

COdataassimilationisnowbasedonbothMOPITTandIASIsensorresults,thelatterbeingnovelinMACC.

IASI is an infrared Fourier transform spectrometer developed jointly by CNES (the French spatialagency),andbyEUMETSAT.IASIismountedon-boardtheEuropeanpolar-orbitingMetOpsatellite.It contributes to atmospheric composition measurements for climate and chemistry applicationswith high horizontal resolution and sampling, and with 1 km vertical resolution (Clerbaux et al.,2009).Toreachthisobjective,IASImeasurestheinfraredradiationoftheEarth’ssurfaceandoftheatmospherebetween645and2760cm-1atnadirandalonga2200kmswathperpendiculartothesatellitetrack.Atotalof120viewsarecollectedovertheswath,dividedas30arraysof4individualField-of-viewsvaryinginsizefrom36 ⋅πkm2atnadir(circular12kmdiameterpixel)to10 ⋅20 ⋅πkm2atthe largerviewingangle. IASIoffers inthisstandardobservingmodeglobalcoveragetwicedaily,withoverpasstimesataround9:30and21:30meanlocalsolartime.

An observations operator for FORLI (Fast Optimal Retrievals on Layers for IASI) (Hurtmans et al.,2012)COdataanditsadjointhasbeendevelopedandwasintegratedintheEURAD-IMassimilationsystem.Inafirststep,theobservationoperatorcomputesCOpartialcolumnsfor19layers.Levels1to18correspondto18atmosphericlayerswith1kmthickness,level19tothelayerfrom18to60km.Finally,theFORLIaveragingkernelvectorforthepartialcolumnsin19verticallayersisappliedtocomputethemodelequivalent. Figure10.2showsresultsfroma3d-varassimilationexperimentof IASI FORLI-COdata forMarch25, 2013. The EURAD-IMbackgroundCO columnsdonot have ageneralbias:atMarch25 theCOcolumncontentwas ratheroverestimated innorth-eastEuropeandunderestimatedinsouth-eastEurope.Thebiaswassignificantlyreducedintheanalysisrun.

Page 102: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page102of148

Figure10.2.AtmosphericCOcolumnsforMarch25,2013assimilationinEURAD-IM.Upperleft:backgroundmodelequivalents,upperright:IASIFORLI-COdata,lowerleft:backgroundminusmeasurements,lowerright:analysisminusmeasurements.

10.1.3IASIOzonedataassimilationmodule

IASI datawas integrated in the data stream, and related observation operatorwas developed orupdated,andusedtoupgradefreetroposphericozonevalues.CNRS-LISAhascontinuedassimilationof IASI ozone observations, to consolidate and optimize assimilation work performed already inMACC-I.Beyondthis,CNRS-LISAprovidedaIASIozoneproduct,forexamplepartialtropospheric0-6kmozonecolumns,over theGEMS–MACCarea forassimilation in regionalCTM’s.Theaveragingkernel,majorpartoftheobservationoperator,ispartoftheproduct(Eremenkoetal.,2008,Dufour,2009,Zyryanovetal.,2010).OtherpartnersusedtheFORLIorSOFRIDozoneproducts.The entire set-up of the LISA assimilation system (CHIMERE-EnKF) based on an EnKF approach isdescribed inComanetal (2012).Theprojectionof themodelsimulations inobservationspaceviatheobservationoperatorismadeinordertocalculateinnovations.Thetropospheric0-6kmozonecolumnsfromIASIinstrumentarederivedfollowingtheEremenkoetal(2008)approach;averagingkernels coming from the retrievals are used to correct the simulated profile where the verticalsensitivityoftheobservationsarethehighest.Observationserrors(diagonalelements)arecomingfromtheretrievalprocessandaGaussianmodel isusedtofulfil thenon-diagonalelementsoftheerrorcovariancematrixRtakingintoaccounttheerrorcorrelation.).Threeassimilationexperimentshave been conducted for 1-month periods in summer 2007, 2009 and 2010. Results of 2007experimentarepresentedinComanetal(2012),resultsofthe2010experimentarepresentedinthecase study ofMACC-III. Here after (Figure 10.3) a comparison between themean ozone columnsbefore assimilation (left panel), the IASI data (central panel) and the mean ozone columns after

Page 103: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page103of148

assimilation(rightpanel)forthe13July2010usinganensembleof20membersand30assimilatedpixels.

Figure10.3.0–6kmozonepartialcolumns(DobsonUnits)fromthemeanoftheforecastensemble(i.e.beforeassimilation;leftpanel),IASI(centralpanel)andthemeanofanalysisensemble(i.e.theensembleafterassimilation;rightpanel)forthe13July2010.(colorscalevariesfromblack,representingrelativelylowvalues,topinkrepresentingrelativelyhighvalues).

10.1.4MODIS,IAODdataassimilationmodule

AODfromMODIS,SEVIRI,wasintegratedinthedatastreamandusedtoupdateparticulatemattervalues.TherelatedAODobservationoperatoroftheaerosolmoduleofthechemistrymodel(anditsadjoint)weredevelopedandintegratedintheDAalgorithm.The observation operator and its adjointwere applied onMODISAODmeasurements in a 3d-varassimilationexperiment. Figure10.4 showsmodel equivalentsof theaerosoloptical depthat 550nmforMarch24,2013.TheassimilationhasstronglyimprovedhighAODvaluesmeasuredovertheMediterranean Sea southof Italy,whichmost probably originate froma Saharadust outbreak. Insomearease.g.overtheBalticSeaAODvaluesareoverestimatedintheanalysis.

Page 104: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page104of148

Figure10.4.AODat550nmforMarch24,2013inEURAD-IM.Upperleft:backgroundmodelequivalent,upperright:MODISmeasurements,lowerleft:analysismodelequivalent,lowerright:analysisminusbackground.

Another similar example is presented on Figure 10.5 for the MOCAGE-Valentina system, showing a clearimpactoftheassimilationofMODISAODdataandmuchimprovedcomparisonwithindependentobservations(fromMSG/SEVIRI).

Figure10.5.AODforthe29thofJune2012(upperleft)forecastedbyMOCAGE,(upperright)MODISobservations,(lowerleft)withassimilationofMODISAODinMOCAGE-Valentina(assimilationcyclesstartedonthe15thofJune2012)and(lowerright)fromMSG/SEVIRIobservations(independentmeasurements).

Page 105: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page105of148

10.2Assimilationalgorithmextension

This work package introduced a couple of improvements of the assimilation algorithm, whichsubstantiallyextendsitsapplicability,whileimprovingrobustness,andstability.Theworkismadeintheframeofmakingminorimprovementstoexistingproductionsystemstomaintainperformance.Thefinaltaskinvestigatedperformanceproblemsarisingfrompreoperationalexperiences.

10.2.1Aerosoldataassimilation

Theaerosoldataassimilationwaseither introducedorrefinedtoaddressthefollowing issues:theObservation-minus-modeldifferenceinAODwasadjustednotsimplybyasuitablefactorappliedtoall components. Rather, the adjustment is introduced according to theheight dependent forecasterrorvariances.Furtheraregularisationoftheaerosolcomponentsis introducedbyaconstrainingfunctiontoassurerealisticportionsofaerosolcomponents.During MACC-III, Met.No increased the computational efficiency of the assimilation chainsignificantly,achievingan80%reduction inwall time,andallowinganon-timedeliveryofanalysiswithNO2andO3assimilation.ThiswasmadepossiblebyusinganewFFTlibraryandparallilizationofthecode.

10.2.2Biascorrectionscheme

Thebiascorrectionschemeaccountsfortypical(probablytimedependent)modelbiasesasfoundbyownevaluation,orEVAandENSevaluation.

Bias correction schemes based on different statistical and probabilistic techniques have beenimplemented. First tests have shown promising results, however the schemes need continuousevaluationatleastaftermajormodelupgrades.

10.2.3Dynamiccovarianceevolution

This task focused on the optimal exploitation of background/forecast information from ensemblerunsand the related formulationascovariances.The forecast/backgrounderrorcovariancematrixhasbeenmadeflow-dependent.Thishasbeendoneeitherbyapplicationofthe“NMC”-methodorbyexploitationoftheENSmodelensemble,orbyexploitationofthemodelensembleusedfortheEnsemble Kalman filter. The choice is based on individual partners’ assimilation algorithmconfiguration.

Page 106: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page106of148

10.2.44D-varemissioninversion

AprototypestudyaimedatoptimisingjointlyofCO2initialvaluesandanthropogenicandbiogenicCO2fluxeswithEURAD-IMforaclosedCO2budgetoftheRurcatchmentarea.OnecentralstepistheinitialisationofoccurringCO2fluxes.Lateralboundaryfluxesarededucedbyanestingapproachfrom15to5to1kmhorizontalresolution,whereboundaryfluxesforthemotherdomain(15km)areusedfromtheMACCIIIproject.AnthropogenicemissionsareusedfromtheTNOinventory, supplying point and area sources of CO2 split into 10 SNAP codes for Europe for thereferenceyear2005.Areasourceshadtobedown-scaledto1kmresolution(usingaGIS).The biogenic fluxes are calculatedwithWRF (Version 3.6.1) using CLM4.0 as land surfacemodel.While photosynthesis is already included in this configuration leaf respiration is implementedaccordingtoCollatzu.a. [1991]andsoilrespiration is implementedasanArrheniustypeequation(Eq. 8 of Lloyd u. Taylor [1994]). The calculation of photosynthesis and leaf respiration uses theaveragedmonthlyMODIS LAI givenwithWRF (see Figure 10.6). To improve themodelling of thepassivetracerCO2anewadvectionschemeanditsadjoint,theabsolutemonotoneWalcekschemewasimplementedintheEURAD-IM.ThisschemepreventsspuriouswigglesoccurringatsharpspatialgradientsofCO2concentrations(especiallyclosetostronganthropogenicpointsources).Duetothelow numerical diffusion of the Walcek scheme, many structures of anthropogenic sources andbiogenicfluxesareresolvedinthelowestlayeroftheatmosphericCO2concentration.The 4D-var assimilation scheme was extended to optimise the emission factors of two fluxes(anthropogenic and biogenic), assuming the diurnal circle is well captured by the TNO emissionsinventoryand themodelledbiogenic fluxes. It isnoweffectivelypossible tooptimise initial valuesandtheemissionfactorsjointly.

Figure10.6.Biogenicfluxestimation.Allfuxesgivenin[_molCO2m2s]at23July201213UTCat5kmhorizontalresolution.Fromlefttoright:photosynthesis,leafrespiration,soilrespiration.

Page 107: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page107of148

Basedon resultsobtained in thepreviousprojectMACC II, theemissiondata inversionprocedurerestingontheEURAD-IM4D-varschemehasbeenextendedtoidentifyemissioncorrectionfactors,accordingtoElbernetal.(2007.Emissionfactorstobeimprovedcompriseobservedspeciesinthefirstplace,butalsonotobserved,yetinfluencedbychemicalcoupling.

As a control and in order to account for fine scale structures, a nesting technique with highhorizontalresolutionwasapplied,startingwithEuropeasmotherdomain(15kmresolution),goingdown to the first nest covering Central Europe (5 km resolution) and ending with the area ofNorthrhine-Westfalia(1kmresolution).

Both ground-based observations of NO2, NO, SO2, CO and O3 (from Airbase) and satelliteobservationsofO3(IASI),NO2(OMI)andCOMOPITT)wereassimilatedfortheseexperiments(seeFigure10.7).

Figure10.7.Topline:EmissioncorrectionfactorsinferredbyEURAD-IMforthephotochemistryreferencecaseJuly2010,givenTNOemissionsforaworkingdayasbackgroundvalues.EmissioncorrectionfactorsforNO2(left),SO2(middle),andCO(right).Inmostareas,anincreaseofemissionratesisanalysed.Bottomline::analysisminusbackgrounddifferencesforsurfacelayerO3concentrationsinthe15kmresolutionEuropeandomain,duringnoon,atthefirst(left)andthefifth(right)dayoftheozoneepisode.

Application of the 4d-varmethod for several days in a row lead to optimized initial data for thefollowing day and so to an accurate analysis state for it. For instance, the fifth day of the ozoneepisode achieves higher reduction of the cost function than the first day. Figure 10.7 showsdifferencesbetweentheanalysisandthebackwardrunforthefirstandthefifthdayoftheozoneepisode,forsurfaceO3concentrations.AfterfiveassimilationdaysthereisasignificantcorrectionoftheNO2emissionfactors,asexpected.

Page 108: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page108of148

10.3Aposteriorivalidation

In this workpackage, state-of-the-art methods for a posteriori validation of data assimilationalgorithms were applied. On the basis that all partners used some type of best linear unbiasedestimatorasdataassimilationalgorithm,itmustbenoted,thatonlyleastsquaretypeoptimacanbeconsidered. The main validation activity made use of χ2-validation with spatial and temporaladjustments.Furtheron,casestudieswithNRTdatawithheldforcontrolwereperformed.Allworkinthispackagewastoguaranteethequalityofdeliverablesandperformanceoftheindividualdataassimilationsystems.

10.3.1χ2-validationregionallyandseasonallyresolved

Models performeddifferentlywell in different regions subject to emission impact and seasons orweather conditions. Likewise, the observation representativeness varied. The χ2−testing activitiesensure a proper balance between the observation and forecast error covariance and are used tovalidatethemutualconsistencyofbackground(forecast)errorcovariancematricesandobservationerrorcovariancematrices,therebyonlybeingabletoproofanecessary,yetnotsufficientcondition.

10.3.2Casestudyanalyseswithnon-NRTdata

InMACCIII,thetwotestepisodesselectedinMACCIIwereextendedtocoverafullmontheach:afairlymoderateaestivalphotochemicalstandardcaseandahibernalcasewithsignificantlyelevatedaerosol concentrationswere analysedwith theprototypeoperational data assimilation algorithm.FollowingasurveyofepisodesduringtheMACCera,thetimespanof07.07.2010-05.08.2010wastakenfortheformer,andthetimespan15.01.2012–15.02.2012forthelatter.

Asforthedata,anexpandedvalidateddataset,theobservationsofwhicharenotavailableinNRT,is taken foramorecomprehensive control. Thesedata compriseEBASdata from theACTRISdataserver, and WOUDC (World Ozone and Ultraviolet radiation Data Centre), to ensure a thoroughQA/QCvalidation.Whiletheepisodeselectionwassolelydevotedtosuitableatmosphericchemistryconditions,noexternalcasestudydatawereavailable,aswaswithIAGOSdata.

10.4Mineraldust,volcanoandfiredataassimilation

This work package focused on the development of measures to address data assimilationmodificationsforspecialtransientatmosphericconditions.These includebiomassburning,mineraldust events, and, based on recent experiences in Europe through the Eyjafjöll eruption, volcanicemissions. This work package engages in developing suitable solutions and upgrades tomaintainproductdeliverywhenexceptionaleventsoccur.

Developmentswerecarriedoutinthreemainareas:

• awildfiredataassimilationmodulewascompletedbyRIUUK.Itincludesanextendeddataassimilationalgorithmtoidentifythechemistrymodels’changetowardlocalfireemissions,estimatetheforecasterrors,andperformthedataassimilationprocedure.Tocomplywiththe global models’ system, biomass burning emissions calculated by the Global Fire

Page 109: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page109of148

AssimilationSystem(GFASV1.0)havebeenimplementedintheEURAD-IMdataassimilationsystem.TheassimilationofGFASdatahasbeensuccessfullytestedonthe2010Russianpeatfires.

• amineral dust data assimilationwas developed. It identifies of chemistrymodels’ changetoward mineral dust load, dust controlled lateral boundary values obtained from globalmodels(sub-projectAER),andperformanceofthedataassimilationprocedure.

• the data assimilation algorithmswere extended to identify the chemistrymodels’ changetoward volcanic emission load and, if applicable, volcanic emission controlled lateralboundary values obtained from global models, estimate the forecast errors, adapt thecovarianceformulation,andperformthedataassimilationprocedure.

10.5References

Biermann, U.M., B.P. Luo, and T. Peter, Absorption spectra and optical constants of binary andternarysolutionsofH2SO4,HNO3,andH2Ointhemidinfraredatatmospherictemperatures,J.Phys.Chem.A,104,783-793,2000.

Coman,A.,G.Foret,MatthiasBeekmann,MaximeEremenko,G.Dufour,etal..Assimilationof IASIpartial tropospheric columns with an Ensemble Kalman Filter over Europe. AtmosphericChemistryandPhysics,EuropeanGeosciencesUnion(EGU),2012,12(5),pp.2513-2532.

Clerbaux, C., A. Boynard, L. Clarisse, M. George, J. Hadji-Lazaro, H. Herbin, D. Hurtmans, M.Pommier, A. Razavi, S. Turquety, C. Wespes, P.-F. Coheur: Monitoring of atmosphericcompositionusingthethermalinfraredIASI/MetOpsounder,Atmos.Chem.Phys.,9,6041-6054,2009,doi:10.5194/acp-9-6041-2009.

Dee, D. P. , Bias and data assimilation. Q.J.R. Meteorol. Soc., 131: 3323–3343. doi:10.1256/qj.05.137,2005.

Dee,D. P. andUppala, S. (2009), Variational bias correction of satellite radiance data in the ERA-Interimreanalysis.Q.J.R.Meteorol.Soc.,135:1830–1841.doi:10.1002/qj.493.

Dufour,G.etal.:IASIobservationsofseasonalandday-to-dayvariationsoftroposphericozoneoverthreehighlypopulatedareasofChina:Beijing,Shanghai,andHongKong,Atmos.Chem.Phys.,10,3787-3801,doi:10.5194/acp-10-3787-2010,2010.

Dufour,G.etal.,ValidationofthreedifferentscientificozoneproductsretrievedfromIASIspectrausingozonesondes.AtmosphericMeasurementTechniques,5(3),611–630.doi:10.5194/amt-5-611-2012,2012.

Evans, T.N. and G.R. Fournier, Simple approximation to extinction efficiency valid over all sizeparameters,AppliedOptics,29,4666-4670,1990.

Eremenko,M.etal. (2008),TroposphericozonedistributionsoverEuropeduringtheheatwave inJuly 2007 observed from infrared nadir spectra recorded by IASI, Geophys. Res. Lett., 35,L18805,doi:10.1029/2008GL034803.

Elbern H, A. Strunk, O. Talagrand: Emission rate and chemical state estimation by 4-dimensionalvariationalinversion,Atmos.Chem.Phys.,7,3749-3769,2007.

Gaubert,B.etal.:Regional scaleozonedataassimilationusinganensembleKalman filterand theCHIMERE chemical transportmodel,Geosci.ModelDev., 7, 283-302,doi:10.5194/gmd-7-283-2014,2014.

Page 110: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page110of148

Page 111: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page111of148

11.Europeanensembleair-qualityanalysesandforecasts(ENS)

The ENS sub-project primarily focuses on the delivery and the verification of the prototypeoperational European-scale regional Nearly Real Time (NRT) air-quality services. In directcontinuation from MACC, this service is based upon an ensemble of forecasts and analysesperformed at seven centers in Europe, recognized for their continuing experience in providingroutine operational or pre-operational forecasts at the scale of Europe for several years. Theassimilation and forecast suites employed inMACC-II are CHIMERE (operated by INERIS, France),EMEP (operated by MET.NO, Norway), EURAD-IM (operated by RIUUK, Germany), LOTOS-EUROS(operatedbyKNMI,Netherlands),MATCH(operatedbySMHI,Sweden),MOCAGE(operatedbyMF-CNRM,France)andSILAM(operatedbyFMI,Finland).

11.1Regionalairqualityproduction

Everyday,forecastsforeachhourandanalysesforeachhourforthedaybeforeareproducedbythesevendesignatedcenters.Theresultsoftheseven individual forecastsandanalysesareprocessedcentrallyatMF-CNRMbycomputingensembleproductsona0.1°latitudex0.1°longitudegridandaseries of verification products, and by displaying the regional NRT products on theMACC-II webplatform.

The individual forecasts/analyses make use ECMWF IFS operational forecast data for theirmeteorologicalforcings,anthropogenicandfireemissionsfromEMISandFIRsubprojects,aswellasthechemicalboundaryconditionsprovidedbyGDAsubprojectforchemicalcompoundsandbyAERforaerosols.Thesevenregionalairqualityassimilationsystemsproducing theanalyseshavebeenfurtherdevelopedduringMACC-IIandMACC-IIIbasedontheresearchworkdoneinEDAsubproject,inparticularonthecombineduseofsurfaceandsatellite-basedobservations.

WithinMACC-II project, the following extensionsof the regional product portfoliowere achieved,basedontheuser’srecommendations:

• The European domain has been enlarged from 15°W-35°E 35°N-70°N to 25°W-45°E 30°N-70°N. It now fully covers a large European continent. This extension was put in place inNovember2012.

• The 72h forecasts run inMACC have been extended to 96h inMACC-II, still with hourlyoutputs.ThisextensionwasalsoputinplaceinNovember2012.

• Based on the MACC developments, birch pollen 96h forecasts at surface have beenproduceddailysincethe2013season(from1stofMarchto30thofJune).

• Thenumberofverticallevelsforthedailyforecastshasbeenincreased:additionallytothelevels provided in MACC (surface, 500m, 1000m and 3000m altitude), the 50m, 250m,2000mand5000mareprovidedsinceMay2014.

• Thenumberofspeciesprovidedforthedailyforecasthasalsobeenincreased:additionallyto the MACC core species (ozone, NO2, SO2, PM2.5, PM10, CO), the following additionalspeciesareproducedsinceJune2014:NO,NH3,totalNMVOC(Non-MethaneVolatilOrganicCompounds),PANs(PAN+PANprecursors).

Theselasttwoextensionsaremainlydesignedforusersrunningairqualityforecastmodelsatfinescale. Thechoiceof thenumberofadded levelsandadded species is a compromisebetween theusers’requirementsandanaffordableproductionsystem.

Page 112: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page112of148

DuringMACC-IIIproject,theensembleanalysisofNO2atsurfaceisnowproducedadditionallytotheensembleanalysisofO3. Thiswasmadepossible thanks to the increaseof thenumberofmodelsmakingtheassimilationofNO2.

Thewhole setofextensionsdescribedabove increases largely thedata transfervolumeand time,theprocessingtimeandtheassociatedstorage.Thisiswhytechnicalworkwasneededtooptimizethe seven individual production times, the data format and the ensemble processing to makeforecastdataavailableinthemorningfollowingMACC-IIuser’srequirements(nowat07UTCforthefirst48hoftheforecast).

For the daily analysis production, the seven systems assimilatemainly the observations from theEuropeanairqualitymonitoringstations.Thus,thedeliverytimeoftheanalysesistighttothetimeof availability of these observations. At the beginning ofMACC-II, these datawere acquired as inMACC, country by country, based on bilateral agreements. An important piece ofwork has beendone in collaborationwithOBS subproject to prepare the use of the EEA (European EnvironmentAgency) NRT database instead of the country-by-country system. The EEA database gathersEuropeanairqualitydatainNRT.MF-CNRMnowacquiresonadailybasisthesedatafromEEA,storethemintheMétéo-Franceoperationaldatabaseandusethemforpreparingacommondatasetforassimilation in the seven individual models. Currently, the EEA does not obtain and thus cannotprovidetheairqualitydataearlyenoughtoallowthesevencenterstoproduceforecastsstartingonthepreviousdayanalysis.

11.2Verificationofregionalairqualityproducts

TheEEANRTdatabase is alsoused for the calculationsof the verificationproducts insteadof thedata gathered country by country as done during MACC project. The verification procedures inMACC-IIIareindirectcontinuationofMACC-II.Theyincludetheproductionofmapsandofstatisticalindicatorsonadailybasisdisplayedontheregionalwebsiteand,additionally,6-monthlydossiersforthe seven individualmodels and the ensemble evaluating the seasonal (3months) performances.The set of statistical indicators has been enlarged in MACC-III following VAL subprojectrecommendations.Itnowincludesthenormalizedmodifiedmeanbiasandthefractionalgrosserror,which are complementary to themean bias, the root-mean square error and the correlation. Animportantevolutionoftheverificationprocedurefortheforecast istheselectionoftheairqualitymonitoringstationstobeusedinthecalculationofthestatisticalindicators.Thisselectionisdonetotake into account the typology of measurement sites because there is no uniform and reliablemetadatacurrentlyforallregionsandcountries.ThedataselectionfollowstheworkthathasbeencarriedoutinMACC(JolyandPeuch,Atmos.Env.,2012)tobuildanobjectiveclassificationofsites,basedon thevalidatedpastmeasurementsavailable inAirbase (EEA).This classification isused inordertorestrictverificationtothesitesthathaveasufficientspatialrepresentativenesswithrespectto the model resolution (10-20 km). The statistical approach using only representative sites -accordingtotheobjectiveclassification-isclearlythewayforward(asitdoesnotalsothintoomuchtheNRTdataavailable),leadingtoageneralsignificantimprovementoftheoverallskillscores.Fortheforecasts,allthedataselectedareusedfortheverification.Fortheanalysis,about20%ofthesedataarenotusedintheassimilationbutkeptfortheverification.Thepost-processingofverificationoftheanalysishasbeendevelopedinthecourseofMACC-IIIbutisnotyetoperational.

To complement this evaluation, specialworkwasdedicatedon theMediterraneanarea, using forreferencetwomodelsrunbyMediterraneanpartners(AEMETandAUTH)athigherresolutionovertwodomainsofinterest(SpainandGreece).Thiswasinviewofimprovingtheskillsofregionalcore

Page 113: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page113of148

servicemodels in this specificarea,which is recognizedaschallenginggiven thespecificitiesof itsdynamicsandchemistry.Thisworkisnotintendedtobeoperationalbuttodocumentinaresearchapproachthestrengthandweaknessesofthesevenmodelsandtheensemble.TheAEMETandAUTmodel high resolution forecast maps are available daily on the MACC-II website. Additionally,semestrial reports have been provided including detailed analyses of case studies for the highresolution models and the ensemble, where/when higher resolution matters. This analysis wasextendedtothesevenindividualmodelssincethelasttworeportsofMACC-II.

Thescoresoftheensembleforozoneforthe latestsummerandforPM10forthe latestwinter inMACC-II/MACC-IIareshowninFigure11.1and11.2,respectively.

Figure11.1Ensembleozonerootmeansquareerrorinµm/m3(leftpanel)andcorrelation(rightpanel)forsummer2014(June,JulyAugust)asafunctionoftheforecasttime.

Figure11.2.EnsemblePM10rootmeansquareerrorinµm/m3(leftpanel)andcorrelation(rightpanel)forwinter2014/2015(December,January,February)asafunctionoftheforecasttime.

An importantresearchefforthasbeendevoted inENSto improvethemodelingof theaerosols inview of better PM10 and PM2.5 forecasts and analyses. Each of the seven individual modelinggroups has upgraded their model versions running for MACC-II/MACC-III with more detailedrepresentations of the different aerosol components. Moreover, the anthropogenic inventoryprovidedbyTNOfor2009inthecourseofMACC-IIwasimplemented.ThelatestTNOinventory(for2011)providedclosetotheendofMACC-IIIwillbeimplementedafterMACC-III.

Pollen modeling and forecast were also one of the important research activities in the ENSsubproject. As part of the extension of the daily forecast to birch pollen since 2013 season, anevaluationofthe2013production(individualmodelsandensemble)hasbeenconductedbasedonaone-year agreement that has been negotiated with 6 EAN (European Aeroallergen Network)countries:Austria,Estonia,Germany,Finland,France,andUkraine,-foroperationalaccesstotheirdailybirchpollenobservations.Thisevaluationhasshownthattheseasonstartisamongthebest-

Page 114: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page114of148

predictedquantities.Seasonendingismuchmoredifficulttopredict.Aknownissuewiththecurrentbirchsourcetermisitsvariability,whichisstilldifficulttopredict.TheresultsofthisstudyhavebeenpublishedinSofievetal.(2015).Inparallel,FMIhasdevelopedduringrecentyearsinco-operationwith EAN the primary biogenic aerosol particle emissions routines to take into account newimportantallergenictaxa:grass,oliveandambrosia(Figure11.3).Themodelingofthesourcetermsfor these three pollen species have been made available to MACC-II regional models. Itsimplementationinthe7modelswasdonewithinMACC-III. Evaluationonpastseasonsisrequiredbeforepossiblefutureextensionsoftheserviceportfoliotothesenewpollens.

Inparallel,anewresearchactivityhasbeenstartedinMACC-IIandcontinuedinMACC-IIIinordertodevelopaCO2modelingcapabilityinthesevenindividualmodels,bycouplingsystemswithsurfacemodels representing natural CO2 fluxes and by taking into account anthropogenic CO2 emissions.Thisisafirststeptowardscarryingoutfuturehigh-resolutionfluxinversions.Allthesevenindividualmodelshavenotreachedthesame levelofdevelopmentsontheCO2modelingactivity.Thisworkneedstobeconsolidatedinthefuture.

Figure11.3.Examplesofpollenconcentration(ingrainsperm3)forecastedbySILAMmodelforbirch(topleft),olive(topright),grass(bottomleft)andambrosia(bottomright).

Theforecastproductsmostusedaretheensembleproducts.This isbecause,byessence,theyaregenerallybetterthantheindividualmodels,althoughtheindividualforecastcanbebetterthantheensemble at some locations and for some dates/times. The method currently used to build theensemble forboththe forecastsandtheanalyses is themedian. Itsmainadvantagesare that it isrobustbecausenotsensibletooutliersand it iscomputationallyefficientsince itonlyrequirestheseven individualmodel outputs. Tests have beenmade to assess the robustness of the ensemble

Page 115: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page115of148

when one ormoremodels are randomly not available. The differences aremainly on root-meansquareerrorsandcorrelations.ResultsinFigure11.4showthat7or6modelsaregivingverysimilarscoresfortheensemble.Using5and4modelsgivessimilarscoresbutisdegradedwithrespecttoensemblesof6or7models.AlsoinMACC,afewalternativemethodstothemedianwereexplored.ThisworkhasbeencontinuedinMACC-IItakingadvantageofnewlypublishedapproachesproposingensemblemethodsforairqualityforecasts.Thefirstresultsofthecomparisonforozoneinsummerhave shown that the median method is generally the best method during daytime, but othermethodshavebetterperformancesatnight.Thisworkneeds tobe testedon longer timeperiodsandotheralternativemethodsneedtobeinvestigated.

Figure11.4.Rootmeansquareerror(toppanel)andcorrelation(bottompanel)forozoneasafunctionoftheforecasttimeinhourforanensembleof7,6,5,4and3modelscomparedtothehourlysurfacestationmeasurementsavailablefortheperiodfromthe1stofJune2014at00UTCtothe1stofSeptember2014at00UTCovertheMACC-IIEuropeandomain.

TheMACC-IIIservice is inthecontinuityofMACCbut includestheorganizationofthetransitionfromprototypeoperationalservicestotheoperationalphase(CAMS).Itisimportanttonotethatthefundsavailable in MACC-II and MACC-III did not provide the possibility of a 7 days/7 days and 24h/24hmonitoring. Consequently, the current system has to be regarded as the demonstrator of the fullyoperationalsystemthatisplannedfortheCopernicusatmosphereservicephase.TheNRTregionalair-qualityservicesaredistributedoverthesevenproductioncentersprovidingrobustnessandrichnesstothe system, but requiring to organise the operations in a distributed context.Météo-France plays acentral role in theEuropeanairqualityproduction since it centralizes the sevenmodel forecastsandanalyses and the observations used for the assimilation, it produces the ensemble forecasts andanalyses, the plots for thewebsite, it develops/maintains theweb site, it produces the files for thenumerical data server and it develops/maintains the ftp numerical data server.Workhas beendonewithinMACC-IIandMACC-IIItoconsolidatetheindividualandcentralchainstoincreasetheirreliabilitybasedonoperationalprocedures,softwareanddatabasessimilartothoseusedforNumericalWeatherPrediction. An improved version of the MACC regional web site has been developed andoperationalised during MACC-III. A prototype data server Inspire compliant (DCPC –like) has been

Page 116: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page116of148

developedto improvetheaccesstoallnumericaldataduringMACC-III.At theendofMACC-III, thereareonlyafewcomponentsthatarenotfullyoperationallinkedtothelackofoperationalfundingsbuttheyhavebeen fullyprepared for theCAMSphase.The feasibilityofaservicecertificationunder ISO9001standardhasalsobeeninvestigated.

All theworkachievedinENSsubprojecthasrequiredan importanteffortofcoordinationbetweentheENSpartnersbutalsowithmanyotherMACC-IIIsubprojects,inparticularwiththeotherregionalsubprojects EDA and EVA (with regular meetings). Moreover, attention was given to fulfill, aspossible,theusers’needsthankstocollaborationswithINTandPOLsubprojects.DuringthecourseofMACC-III, thenumberofdaily/NRTusersoftheregionalproductshas largely increased(above200) together with the variety of applications based on the ENS products (research, operational,institutionalandcommercial).

Thedescriptionof thecurrentstatusof theENSproductionchains is thesubjectofapaper in theMACCspecialissue(Marécaletal.,GMD,2015).

11.3References

Marécal.,V.,V.-H.Peuch,C.Andersson , S.Andersson, J.Arteta,M.Beekmann,A. Benedictow,R.Bergström,B.Bessagnet,A.,Cansado,F.Chéroux,A.Colette,A.Coman,R.L.Curier,H.A.C.DeniervanderGon,ADrouin,H.Elbern,E.Emili,R.J.Engelen,H.J.Eskes,G.Foret,E.Friese,M.Gauss,C.Giannaros,J.Guth,M.Joly,E.Jaumouillé,B.Josse,N.Kadygrov,J.W.Kaiser,K.Krajsek, J. Kuenen, U. Kumar, N. Liora, E. Lopez, L. Malherbe, I. Martinez, D. Melas, F.Meleux,L.Menut,P.Moinat,T.Morales,J.Parmentier,A.Piacentini,M.Plu,A.Poupkou,S.Queguiner, L. Robertson, L. Rouïl, M. Schaap, A. Segers, M. Sofiev, M. Thomas , R.Timmermans,Á.Valdebenito,P.vanVelthoven,R.vanVersendaal,J.Vira,A.Ung,AregionalairqualityforecastingsystemoverEurope:theMACC-IIdailyensembleproductionGeosci.Mod.Dev.,8,2777-2813,doi:10.5194/gmd-8-2777-2015,2015.

Sofiev.M.,U.Berger,M.Prank,J.Vira,J.Arteta,J.Belmonte,K.-C.Bergmann,F.Cheroux,H.Elbern,E.Friese,C.Galan,R.Gehrig,R.Kranenburg,V.Marécal,F.Meleux,A.-M.Pessi,L.Robertson,O.Ritenberga,V.Rodinkova,A.Saarto,A.Segers,E.Severova,I.Sauliene,B.M.Steensen,E.Teinemaa,M.Thibaudon,V.-H.Peuch,Multi-modelsimulationsofbirchpolleninEuropebyMACCregionalensemble,Atmos.Chem.Phys.,15,8115-8130,2015.

Page 117: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page117of148

12.Validatedairqualityassessment(EVA)

The so-called EVA services relate to detailed analysis of air quality over past situations thanks tovalidatedmaterialissuedfromobservationnetworks,Earthobservationandmodelling.Therefore,aposteriori validated air quality assessments for Europe, based on re-analysed air pollutantconcentrationfieldsareproposed.“Re-analysis”meansthatsimulationsofairqualityareperformedbyregionalchemistry-transportmodelsoverpastyearsand“improved”thankstotheassimilationofavailable validated in-situ and satellite observations. Re-analyses are compiled and commented inthe regional airquality assessment reportswhichdescribe,witha yearly frequency, the stateandtheevolutionofbackgroundconcentrationsofairpollutants inEuropeancountries.Specialcare isgiven to regulatory pollutants characterised by the influence of long range transport, correctlycaughtbyEuropeanscalemodellingsystems:ozone,nitrogendioxide,particulatematter(PM10andPM2.5).Focusonspecificpollutionepisodesthathappenedduringtheyearwillbeconsidered.

TheEVAserviceiscloselylinkedtotheothertworegionalairqualityservices,ENSandEDA:

• The suite of models used in EVA is the same as the one developed and run in the ENSservicestoprovidedailyairqualityforecastsandanalyses.Thesevenmodelsinvolved(andtheirupgradedversions) areexactly the sameand concernCHIMERE (operatedby INERIS,France),EMEP (operatedbyMET.NO,Norway),EURAD-IM (operatedbyRIUUK,Germany),LOTOS-EUROS (operated by KNMI, Netherlands), MATCH (operated by SMHI, Sweden),MOCAGE(operatedbyMF-CNRM,France)andSILAM(operatedbyFMI,Finland).AsforENS,anensemblemodelisbuiltuponthemedianofindividualmodelresults.

• Re-analysesaredataassimilatedmodelresultsperformedbythedata-assimilationmodelingchainsdevelopedbyeachteam in theEDAsub-project.Generally, in-situobservationdataare issued from the AIRBASE database maintained by the European Environment Agency(EEA)togatherobservationsfromallregulatorymonitoringnetworksintheEuropeanUnionare used. Some teams used more observations; especially RIUUK and KNMI whooperationnalyassimilatedsatelliteobservationsfromNO2.

TheEVAservicesarebasedonahighlevelofqualitycontrol.Indeed,allindividualmodels’resultsasensemblemodel’sresultsareevaluatedagainstasetofobservationsthatarenotusedinthedataassimilationprocess.Thesetofobservationdataselectedforverificationandevaluationprocessesisdefinedbeforeeachyearlyre-analysisprocess.

Within MACC-III, two main challenges needed to be solved to implement operationally the EVAservices:

- Development of the operational air quality re-analysis infrastructure by the individualmodelling teamsanddevelopmentofby thecoordinating team inchargeof theensemblecalculation(INERIS)

- Production of the yearly validated assessment reports with a frequency and a delaycompatiblewiththeusers’needs.Nationalagenciesorpolicybodiesinchargeofairqualitymanagement andmonitoring are among themost active users of the service. During theusers‘meetingsandexchangeswehad,theyrequestedpublicationoftheEVAassessmentsas soon as possible to get them for the regulatory reporting process according to the Airquality Directive (2008/50/EC). The activity of the EVA teams essentially focused on thisobjectiveduringtheMACC-IIIproject.

Page 118: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page118of148

12.1Developmentoftheoperationalinfrastructure

EachmodellingteaminvolvedinEVA(sevenregionalairqualitymodels)developedwithinMACCandMACC-IItheoperationalinfrastructurethatprovidesairqualityre-analysesforthepastyears.Alotof work has been done for the EDA and ENS projects as well for data assimilationmethods andmodel parametrisations and set-up (domain, input data...) respectively. At the individual modellevel, EVA production was dependant of progress of work in both EDA and ENS sub-projects.DuringtheMACC-IIIprojectonlyfewdevelopmentsweremadeonthere-analyseschains.Thesamemodelconfigurationswerekeptforrunningthe2013assessmentreports.Workwasmorefocusedon the analysis of the results, especially in termsof performanceof data assimilation systemandchoiceofthedatatobeassimilated.AttheendofMACC-IIIthedataassimilationchainshavefinallybefrozenaccordingtotheset-updescribedinthedeliverablereportD54.1describingtheindividualmodellingplatforms:

• All the seven models assimilate at least two chemical compounds among the speciestargetedbytheservice:ozone,NO2,PM10andPM2.5.

• Ozonedatafromregulatoryreportingareassimilatedbyallthemodels• Two models assimilate in operational way NO2 columns from satellite retrievals (OMI,

GOME)andatleasttwootheroneshaveimplementedtestingmodellingchains.

Howeveritshouldbenotedthatotherdatasets(forinstancethoseprovidedbyresearchairqualityandatmosphericcompositionnetworks)wereusedtoanalyseairpollutionlevelsandepisodesthatoccurredintheyearsstudiedduringtheMACC-IIIperiod.InparticularchemicalcompositionofPMfrom the ACTRIS network and vertical distribution of air pollutant as available from the LIDAREARLINETnetworkswereconsidered.

The term “operational infrastructure” relies not only on the modelling chains, but also on thematerial components (supercomputing, secured spacedisk forarchives...). ThereforeMACC teamshadtosecuresupercomputingsystemstobeabletobeartheEVAproduction,whichrequiresveryhighresources(onefullyearofsimulationsforayearlyreport).

INERISwasresponsibleaboutthere-analysisensemblecomputation. ItcollectedthemodelresultsprovidedbytheEVApartners,runthepost-processingprocedurestobuildupensemblereanalyses,perform an overall assessment of models performances (individual and ensemble) and produceautomatically a set of indicators to be used in the EVA reports. This chain of treatment andevaluation process was developed and implemented by INERIS, and became fully operational in2013.

12.2Assessmentreportproductionandroutineevaluation

BytheendofMACC-IIIprojectthe2013assessmentreportanditsevaluationreportweresupposedto be published based on the 2013 observation data collected from regulatory networks by theEuropeanEnvironmentAgency.Thevalidatedobservationdataare reportedby theEEAeachyearon 30th September the latest for the previous year. And generally the validated air qualityobservations are released by the EEA, after cross-checking, at the beginning of the next year(January-February). TheMACC-III Ensemble validated re-analysis process is basedon this deadlineforthedeliveryoftheAssessmentreportsinJuly(seedescriptionindeliverablereportD54.2).

Page 119: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page119of148

Howevertheyear2014wasspecialbecauseforthefirsttimetheMemberstateshadtoreporttheirvalidated observation data according to the new implementation provisions of the Directive2008/50/CEwhicharedescribedintheDecision2011/850/EC.Datatobereportedareassociatedtoregulatorydatasets characterizedbynew informationandmetadata tobeprovided.New formatscompatiblewiththedeliveryofwebservicesweresupposedtobeadoptedbytheMemberStates.This is the so-called AQ e-reporting process. The data flow associated with the validated AQobservation is called E1 and the AIRBASE database is replaced by the AQ e-reporting database(http://www.eea.europa.eu/data-and-maps/data/aqereporting).

Unfortunately some Member States were not fully ready to report according to the newimplementation provisions in September 2014. Therefore it took more time than usually for theREEAtocollectallthedataandcheckthem.InApril2015,theypublishedtheinterimstatusofthedata collected through the E1 data flow. It is displayed below for information. In April, reportingfrom 14 countries was not completed AQ directive. This explained the reason why the validatedobservationdatahavenotbeenreleasedbeforeMay2015.Thedirectconsequences for the2013validatedassessment report is a consequentdelay compared towhatwasplanned.WhenwritingtheMACC-IIIproposal,weexpectedtobeabletopublishthereportbyJuly2015(withrespectwiththe same time line adopted for the 2012 report). But with the publication of the validatedobservationdata inMay, itwasnotpossible to fulfil thisobjective.Wehaveexceptionally revisedtheprocesstobeabletopublishthe2013assessmentreportbySeptember2015.

Figure12.1.OverviewoftheE1airqualityreportinginApril2015.Onlycountriesindarkgreencompletedthereportingprocess(source:http://www.eionet.europa.eu/dataflows/pdf2014/flow_summary?flow=AQDIPR-E1a).

ThereforetheEVAteamswithinMACC-III focusedontheproductionofan interimreportthatwasrequestedbytheuserstobenefitfrompreliminaryresultsoftheEVAservicesbeforetheregulatoryreportingtime(beforeSeptemberoftheyearY+1fortheyearY).Soweproducedfor2013interimre-analysesbasedonobservationdatareportedtotheEEAaccordingtothe“up-to-date”dataflow(or near real time) for which validation is not mandatory. This data flow is called in the newimplementation provision “E2” and it became mandatory for all the pollutants measured byautomaticdevicesinMarch2015.FortheInterim2013production,weusedavailable“up-to-date”data (so perhaps not exhaustive) and all the teams computed interim re-analyses based on thisdataset.The resultsallowedtobuildup interimEnsemble re-analysesand toelaboratean interimreportbasedonthoseresults.Figures12.2and12.3provideacoupleexamplesofthemaps.

Oncethevalidatedre-analyseswillbeperformed,acomparisonwithinterimresultwillbedoneinordertoaddressqualityissues.

Page 120: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page120of148

Figure12.2.PM10limitvaluefortheannualaverageis40µg/m3.

Figure12.3.T120isthenumberofdayswhen8-hoursaverageexceededthe120µg/m3threshold.

Page 121: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page121of148

12.3DisseminationandCommunication

TheEVAservicesanditsreportshavebeenpresentedduringthe2stMACC-IIIpolicyworkshopheldinMarch2015inVienna.

Throughthewebsite, it is importanttogiveaccessnotonlytothecommentedmapsofairqualityindicators,butalsotothenumericalresults.Aselectionofmaps ispublishedonthePOLwebpage(EVAproductsareconsideredaspolicyrelevantones).AlotofworkhasbeendoneincollaborationofMeteoFrance(CNRM)whocoordinatestheENSsubprojecttopublisharchivesofnumericalEVAdataonaportaldevotedtoregionalairqualityprojectsinMACC-III.For2013,theinterimresultswillbemadeavailableandtheywillbeupdatedwiththevalidatedoneswhenready.

Page 122: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page122of148

Page 123: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page123of148

13.Solarradiationservices(RAD)

MACC-III combines monitoring and forecasting UV services with services for solar energy usersrelying on long-term databases of solar surface irradiance. These services make use of ozone,aerosol and total water vapour content (TWC) results based on global data assimilation andmodellingtogetherwithMeteosatSecondGenerationcapabilitiesforcloudmonitoring.

TheMACC-RAD information system commands the execution of the newHeliosat-4method, andgives access to theMACC-RADdatabase through two services:MACC-RADandMcCleardeliveringtime-series of irradiation. It includes a monitoring of the inputs databases relating to theatmosphereproperties (MACC)and to thecloudproperties (APOLLO),andaqualitycontrolof theconsistencyoftheSSIproducts.TheinfrastructureadoptedfortheMACC-RADinformationsystemisfully aligned with GEOSS (Global Earth Observation System of Systems) recommendations oninteroperability. This ensures a wide dissemination of the results. The service McClear is fullyoperational in a user test phase since January 2014, andMACC-RAD since April 2014. Test phaseoperations have been continued through MACC-III and the service chain has been furtherelaborated.

RegardingthevalidationoftheUVprocessorandprognosticaerosols,recentlyfounddeficienciesintheoperational IFSUVprocessorwere investigatedanda revisedversionof the radiative transferschemehasbeendeveloped,evaluatedandisnowbeingused.

13.1TheHeliosat-4methodandprocessingchain

TheHeliosat-4method isbasedon thedecoupling solutionproposedbyOumbeetal. (2014) thatresults fromMACCandMACC-IIprojects. Itwasdemonstrated that the solar irradianceatgroundlevelcomputedbyaradiativetransfermodelcanbeapproximatedbytheproductoftheirradianceunderclearatmosphereandamodificationfactorduetocloudpropertiesandgroundalbedoonly.Changesinclear-atmospherepropertieshavenegligibleeffectonthelattersothatbothtermscanbecalculatedindependently.

A fast clear-skymodel calledMcClear (Lefèvreet al., 2013)wasdeveloped toestimate thedown-wellingdirectandglobalirradiancesreceivedatgroundlevelunderclearskies.McClearimplementsafullyphysicalmodellingreplacingempiricalrelationsorsimplermodelsusedbefore.Itexploitstherecentresultsonaerosolproperties,andtotalcolumncontentinwatervapourandozoneproducedbytheMACCproject.ItaccuratelyreproducestheirradiancecomputedbythelibRadtranreferenceradiative transfermodelwithacomputational speedapproximately105 timesgreaterbyadoptingtheabaci,or look-uptables,approachcombinedwith interpolationfunctions. It isthereforesuitedforgeostationary satellite retrievalsornumericalweatherpredictionschemeswithmanypixelsorgridpoints,respectively.

Another result fromMACC andMACC-II projects is that the vertical position and the geometricalthicknessofacloudbothhaveaverysmalleffectontheglobalirradiance.Asaconsequence,typicalaltitudes of clouds may be selected instead of updated and localized values. Four types ofclouds -low, medium, high water/mixed phase, and thin cirrus- have been selected as thisinformationisprovidedbytheAPOLLOforeachpixel(3kmatnadir)andevery15minwithamaskcloud-free/cloudyandthecloudopticaldepth.Cloudcoverage,i.e.thefractionofapixelcoveredbyacloud isderived foreachtypeofcloudseparately.Similarly toMcClearclear-sky irradiances, the

Page 124: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page124of148

McCloudcloudyskyirradiancesarecomputedbythemeansofalook-uptableapproachcombinedwithinterpolationfunctionsbetweenthenodesofthetables.

TheHeliosat-4chainimplementstheHeliosat-4method(Figure13.1).Severaldataarereceivedfromvarioussources:MACC,DLRandNASA.Currently,theprocessingforproducingaproductismadeon-request(on-the-fly).

Figure13.1.SchematicrepresentationoftheHelioClim-4chainforcomputingirradianceproducts.ThethreedatabasesarepartoftheHelioClim-4database.Theterm‘near-real-time’hastobeinterpretedaswithin2days.

13.2PreparingforAPOLLONextGeneration

TheHeliosat-4methodwasdevelopedduringMACCandMACC-IIprojects.Oneofthekeyfeaturesisthe separatehandlingof clear skyandcloudysky irradiances. In theoriginalHeliosat-4processingchain, cloud properties are derived by the cloud retrieval scheme APOLLO (AVHRR ProcessingschemeOvercLouds,LandandOcean;Kriebel,SaundersandGesell,1989;Kriebeletal.,2003),analgorithm thatwasdeveloped to infer cloud information fromAVHRR sensorsonNOAA satellites.Slightlyadaptedtotheavailablechannelsandthedifferingviewinggeometry,APOLLOisalsousedtoretrievecloudinformationfromSEVIRI(SpinningEnhancedVisibleandInfraredImager,equippedon Meteosat Second Generation (MSG) satellites) observations. In order to facilitate the changefromSUNSolaristolinuxbasedoperatingsystemsintheprocessingchains,are-codingofAPOLLOwasneeded.Also,thehistoricallygrownAPOLLOcodeisdifficulttomaintainanddoesnotmakeuseofnowadays computerarchitecturewith respect to computational efficiency.Additionally, severalMACCuser’sfeedbackhaspointedoutsomemethoddeficiencieswhenusingtheexistingscheme.

As a consequence from these technical requirements and an envisaged methodologicalimprovement a new APOLLO version has been developed at DLR. This so-called APOLLO_NG(APOLLO_NextGeneration;Klüser,KilliusandGesell,2015) isbasedonthesamephysicalprinciplesas the traditional APOLLO scheme and also uses the AVHRR heritage channels only, just as theoriginalAPOLLO.InthecaseofMSG,thesearethe0.6,0.8,1.6,0.39,10.8,and12.0µmchannels.

Source: MACCParameters: H2O, O3,

aerosol propertiesFrequency: daily

Database atmosphere

Source: DLRParameters: cloud properties

Frequency: 15 min

Database cloud

Source: MINES ParisTechParameters: BRDF model

parameters of ground albedoFrequency: climatology

Database albedo

McClearModel

McCloudModel

Heliosat-4 Method

Manual / automated requestfrom MACC site or SoDa

Service˝Near-real-time˝processing

Page 125: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page125of148

Theinternalclouddetectionprocedurehaschangedfromseveralbinarythresholddecisionstowardsa probabilistic approach, and new cloud property information has been added to the standardoutput.

It isnot intended toprovidean in-depthdescriptionof thealgorithmicdetailshere,but tobrieflyexplaintheprobabilisticconceptbehindAPOLLO_NG(Fig.13.2).IntheoriginalAPOLLOscheme,theclouddetectionofeachpixelisbasedonfivethresholdtests.Theoutcomeofeachtestisassignedwithaspecificadditiveweight.Thenonefinalthresholdisappliedforthedecisionwhetherapixeliscloudyorcloudfree.

The cloud detection in APOLLO_NG still uses the same cloud tests, but in a probabilisticmanner,thustheresultingcloudprobabilityiscomposedoftheprobabilitiesofeachindividualtest.

Figure13.2.Thelinearapproachofdifferentconfidencelevelsandthresholdsusedforthecloudprobabilityestimationfromanobservationofanarbitraryvariablex;Klüser,KilliusandGesell(2015).

Comparisons of APOLLO and APOLLO_NG results are ongoing, preliminary results on cloudmasksand cloud optical depth have been given during MACC-III. A major evaluation against groundmeasurementsofirradiancesisforeseenforwinter2015/16.FirstassessmentsrevealtypicallylargerCOD, which might reduce the positive biases found in MACC-RAD validations and a betterdiscriminationofcloudanddesertsurfacesinearlymorninghoursinwintertime.

13.3TheMACC-RADInformationSystem

ItconstitutestheinterfacetousersandcommandstheexecutionoftheHeliosat-4methodon-the-fly.Throughthe informationsystem(Figure13.3), registeredusers (whichhavesignedtheLicenceterms for MACC Products and Services) may access the two services: Helioclim-4 and McCleardelivering time-series of irradiation. The MACC-RAD Web services 1)are described in existingcatalogues,2)aredeployedontheenergycommunityportal(webservice-energy.org),3)bearclearreferences and links toMACC and the Copernicus Atmospheric Service as awhole, and 4)can beinvokedthroughtheWebusingGEOSS(GlobalEarthObservationSystemofSystems)standards.

Page 126: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page126of148

Figure13.3.SketchoftheMACC-RADInformationSystemanditsintegrationasaGEOSScomponent

DuringMACC-IIIchangeshavebeenmadease.g.hardwarechangesinthesatellitedataflow,datatransfer/handling changes, changes in the web services/user interfaces, activities, and somesoftware re-engineering to improvemaintenance possibilities. Duplications of several parts of theprocessingchainhavebeensetup to increase reliability. Severaluser responsesonquality issueshave been investigated and traced back to bugs in input data handling, abaci selection andatmosphericprofileselectioninMcClearandalsointocloudretrievalissuesovercolddesertsurfacesand for partly cloudy pixels in APOLLO affecting MACC-RAD. Those have been investigated andtreatedifquicklypossible.

Atoolhasbeensetupformonitoringusers’accesstothewebservices.Validandrejectedrequestsare numbered per user, per day, and per summarization type. The access log is updatedautomatically every day and its content can be displayedon awebpage. It can be visualizedperuser,orfortheensembleofthem.

13.4TheMACC-RADandMcClearservices

TheMcClearserviceprovidestime-seriesof irradiationthatshouldbeobservedinanyplaceintheworldatanytimestartingfrom2004-01-01iftheskywereclear,i.e.cloudless.ThisserviceisinplacesinceOctober2012.IthasbeenimprovedintermsofusabilityandisfullyoperationalsinceJanuary2014.

TheMACC-RAD service provides time-series of irradiation at any place in the field of view of theMeteosat satellite, i.e. Europe, Africa, Atlantic Ocean, at any time starting from 2004-02-01. Theservicewasextendedtocovernotonly2013onwardsbutbackto2004duringMACC-III.TheserviceisinplacesinceJanuary2014andisoperationalsinceApril2014.

There are two means of access to each service. One is manual via a web interface located atwww.soda-pro.com (Figure 13.4). The other is for automated access via computers and isdocumentedatwebservice-energy.org.BothMACC-RADandMcCleararecurrentlyinthe‘MACCIII-experimentalroutineproductionmode’.

GEOSSCSR

Request–AnswerRegistered

Operational SystemRadiationDatabase

webservice-energywebservice-energy.org

CatalogOGCCSWINSPIRE-compliant

OGCWebServices• Australia Climate•HelioClim-3• SOLEMI• MACC-RAD

SoDaServicewww.soda-is.org

MACC-RADClient

Page 127: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page127of148

Figure13.4.InterfaceoftheMcClearclientintheSoDaServiceorMACCWebsite.

Followingrequestsofseveralusers,averbosemodeofthe1mintime-serieshasbeenimplementedasadditionaloption(Figures13.5and13.6).Thisprovidesallusedinputparametersandnotjustthesolarirradiancevalues.

Finally,weupdateusagestatisticsforthe2ndquarterof2015(Figures13.7and13.8).

# # Columns: # 1. Observation period (ISO 8601) # 2. TOA. Irradiation on horizontal plane at the top of atmosphere (Wh/m2) # 3. Clear sky GHI. Clear sky global irradiation on horizontal plane at ground level (Wh/m2) # 4. Clear sky BHI. Clear sky beam irradiation on horizontal plane at ground level (Wh/m2) # 5. Clear sky DHI. Clear sky diffuse irradiation on horizontal plane at ground level (Wh/m2) # 6. Clear sky BNI. Clear sky beam irradiation on mobile plane following the sun at normal incidence (Wh/m2) # 7. sza. Solar zenith angle for the middle of the summarization (deg) # 8. atm. Atmospheric profile code: afglus=U.S. standard afglt=tropical afglms=midlatitude summer afglmw=midlatitude winter afglss=subarctic summer afglsw=subarctic winter # 9. tco3. Total column content of ozone (Dobson unit) #10. tcwv. Total column content of water vapour (kg/m2) #11. AOD BC. Partial aerosol optical depth at 550 nm for black carbon #12. AOD DU. Partial aerosol optical depth at 550 nm for dust #13. AOD SS. Partial aerosol optical depth at 550 nm for sea salt #14. AOD OR. Partial aerosol optical depth at 550 nm for organic matter #15. AOD SU. Partial aerosol optical depth at 550 nm for sulphate #16. AOD 550. Aerosol optical depth at 550 nm #17. AOD 1240. Aerosol optical depth at 1240 nm #18. alpha. Angstroem coefficient for aerosol #19. Aerosol type. Type of aerosol: -1=no value 5=urban 7=continental clean 8=continental polluted 9=continental average 10=maritime clean 11= maritime polluted 12=maritime tropical 13=antarctic 14=desert #20. fiso. MODIS-like BRDF parameter fiso #21. fvol. MODIS-like BRDF parameter fvol #22. fgeo. MODIS-like BRDF parameter fgeo #23. albedo. Ground albedo #

Figure13.5.Legendofthecolumnsintheverbosemodeofthe1mintimes-serieseditedthroughtheMcClearclient.

Page 128: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page128of148

# # Columns: # 1. Observation period (ISO 8601) # 2. TOA. Irradiation on horizontal plane at the top of atmosphere (Wh/m2) # 3. Clear sky GHI. Clear sky global irradiation on horizontal plane at ground level (Wh/m2) # 4. Clear sky BHI. Clear sky beam irradiation on horizontal plane at ground level (Wh/m2) # 5. Clear sky DHI. Clear sky diffuse irradiation on horizontal plane at ground level (Wh/m2) # 6. Clear sky BNI. Clear sky beam irradiation on mobile plane following the sun at normal incidence (Wh/m2) # 7. GHI. Global irradiation on horizontal plane at ground level (Wh/m2) # 8. BHI. Beam irradiation on horizontal plane at ground level (Wh/m2) # 9. DHI. Diffuse irradiation on horizontal plane at ground level (Wh/m2) #10. BNI. Beam irradiation on mobile plane following the sun at normal incidence (Wh/m2) #11. Reliability. Proportion of reliable data in the summarization (0-1) #12. sza. Solar zenith angle for the middle of the summarization (deg) #13. atm. Atmospheric profile code: afglus=U.S. standard afglt=tropical afglms=midlatitude summer afglmw=midlatitude winter afglss=subarctic summer afglsw=subarctic winter #14. tco3. Total column content of ozone (Dobson unit) #15. tcwv. Total column content of water vapour (kg/m2) #16. AOD BC. Partial aerosol optical depth at 550 nm for black carbon #17. AOD DU. Partial aerosol optical depth at 550 nm for dust #18. AOD SS. Partial aerosol optical depth at 550 nm for sea salt #19. AOD OR. Partial aerosol optical depth at 550 nm for organic matter #20. AOD SU. Partial aerosol optical depth at 550 nm for sulphate #21. AOD 550. Aerosol optical depth at 550 nm #22. AOD 1240. Aerosol optical depth at 1240 nm #23. alpha. Angstroem coefficient for aerosol #24. Aerosol type. Type of aerosol: -1=no value 5=urban 7=continental clean 8=continental polluted 9=continental average 10=maritime clean 11= maritime polluted 12=maritime tropical 13=antarctic 14=desert #25. fiso. MODIS-like BRDF parameter fiso #26. fvol. MODIS-like BRDF parameter fvol #27. fgeo. MODIS-like BRDF parameter fgeo #28. albedo. Ground albedo #29. Cloud optical depth (value of the nearest acquisition time of the pixel) #30. Cloud coverage of the pixel (percentage from 0 to 100, value of the nearest acquisition time of the pixel) #31. Cloud type (value of the nearest acquisition time of the pixel) -1=no value 0=no clouds 5=low-level cloud 6=medium-level cloud 7=high-level cloud 8=thin cloud

Figure13.6.Legendofthecolumnsintheverbosemodeofthe1mintimes-serieseditedthroughtheMACC-RADclient.

Page 129: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page129of148

Figure13.7.NumberofMACC-RADregistered,activeusersandtheirrequestsperclassofusersin2ndquarterof2015.GeographicalregionsofthecompaniesusingtheMACC-RADservice.

23

39

14

RegisteredactiveMACC-RADusersquarter#22015

Fromcompanies

Researchers/Academics

Policyagencies

Unknowntype

238

366

108

NumberofrequeststoMACC-RADquarter#22015

Fromcompanies

Researchers/Academics

Policyagencies

Unknowntype

15

121 1

RegionofcompaniesusingMACC-RADquarter#22015

EU

Switzerland

USA

SouthAfrica

Brazil

Singapore

Turkey

Page 130: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page130of148

Figure13.8.NumberofMACC-McClearregistered,activeusersandtheirrequestsperclassofusersin2ndquarterof2015.GeographicalregionsofthecompaniesusingtheMACC-McClearservice.

17

24

2 5

RegisteredactiveMcClearusersquarter#22015

Fromcompanies

Researchers/Academics

Policyagencies

Unknowntype

658487

212

NumberofrequeststoMcClearquarter#22015

Fromcompanies

Researchers/Academics

Policyagencies

Unknowntype

9

1

RegionofcompaniesusingMcClearquarter#22015

EU

Switzerland

USA

SouthAfrica

Brazil

Page 131: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page131of148

13.5User’sGuidefortheMACC-RADservice

TheUser’sGuidedocumentstheMACC-RADserviceandtheuser-accessprocedurestothedeliveredproducts.Itisalivingoneandtoevolveastheserviceanditscomponentsevolvethemselves.

PartAdescribesthecommunitiesofusers,theirexpectationsandgivesanoverviewofthecompliance of theMACC-RAD service with those. Part B presents the history of Heliosatmethods, the legacy HelioClim-3 and SOLEMI databases as well as the methods used toconvert satellite images into solar surface irradiance. Part C presents an overview of thenew Heliosat-4 method, of the McClear and McCloud models, and the workflow inHelioClim-4 chain. Part D presents the procedure of quality control of the inputs toHeliosat-4, aswell as the consistency of products estimates. Part E details the deliveringproducts and describes the MACC-RAD information system and its integration as GEOSScomponent.

AnupdatehasbeenprovidedasdeliverablereportD57.5duringMACC-III.

13.6MACC-RADValidationproceduresandreports

Adocumentdefining,describingandexplainingthequarterlyvalidationprocedureforGHIandDNIhas been prepared. A number of stations has been defined in order to monitor them regularly(Figure 13.9). Based on this standard, two consecutive validation reports covering 3months havebeengenerated.

Figure13.9Mapshowingthesevenstationsusedfortheperiodicvalidation.

Page 132: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page132of148

13.7Workshopandtrainingmaterial

MINESParisTech/ArminesandTransvalorhaveco-organisedatrainingcourseinsolarradiationinSophia Antipolis, on 14-16 January 2015. It has gathered 34 engineers in solar energy and onebanker.Most of themwere from European Union (France, Germany, Portugal). A few originatedfromcompaniesinAlgeria,Brazil,MoroccoandTurkey.

Thiswasthethirdeditionofthistrainingcourse.OnemainbenefitofsuchtrainingcoursesliesintheinteractionwithusersoftheservicesofferedbyMACCinradiation:McClearandMACC-RAD.Ontheonehand,participantsmaydiscusswiththeresearchersresponsibleoftheseservicesandwiththeTransvalorcompanythatmanagestheoperationalchainsfortheseservices.Theybetterunderstandtheseservices that theyareusingmoreandmore.Ontheotherhand, researchersandTransvalorbetter understand the concerns of the engineers and may capture their needs. They betterunderstandthedrawbacksoftheservicesormisunderstandingormisuseandplancorrectionsandimprovements.

In contrast with the two previous editions and based on feedbacks from these editions and onconstantinteractionswithusersoftheMACCRADservices,thecontentofthecoursewasbuiltonastoryline throughout the course that deals with the daily problems faced by engineers in solarenergy.ThestorydealswiththeproblemsfacedbyafictionalcharacterPaul,fromthealsofictionalcompany SunnyFlowerworking in solar energy. Paul has been charged by his company to build asolarprojectfromtheverybeginning,fromtheidentificationofthemostpromisingandsuitablesiteregardingthesolartechnology,untilthesupervisionoftheoperationallifeoftheplant.

ThetrainingmaterialofthisclassisavailableasdeliverablereportD56.3ontheMACC-IIIwebsite.

13.8Interactionwithinternationalusercommunities

Severalparticipantsof theMACC-RADteamwereactivewithintheCOSTaction1002on ‘Weatherintelligence for Renewable Energies’. This COST action has ended in October 2014 and severalmembersoftheteamparticipatedinthefinalworkshop.

Also,severalparticipantsoftheMACC-RADteamaretakingpartatworkshopsoftheInternationalEnergyAgency(IEA)Task46‘SolarResourceAssessmentandForecasting’expertnetwork.

ThemeteorologicalcommunityhasbeeninformedaboutMACC-RADresultsinseveralpresentationsattheEuropeanMeteorologicalSocietyAnnualMeetinginSeptember2014.

Thecommunityofscientistsandindustryactinginthefieldofconcentratingsolartechnologieshasbeen informed about MACC-RAD results in presentations at their annual meeting SolarPaces inSeptember2014aswell.

ThesecommunicationshavebeenthebasisforraisingtestusersfortheMACC-RADservicefromthedifferentsolarenergyusercommunities.Obviously, theseactivitiescreatedasufficientnumberoftestusersasreportedabove

13.9EvaluationofECMWFirradianceforecastswithrespecttosolarenergyusers

The European Centre for Medium-Range Weather Forecasts (ECMWF) has recently changed itsoutput variable selection in order to include direct irradiances. Additionally, theMACC near-realtimeservicesprovidedailyanalysisandforecastsofvariousparametersbasede.g.onnewaerosol

Page 133: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page133of148

properties. The operational ECMWF/IFS forecast systemwill profit on themedium term from theMACCaerosolforecasts.

Figure13.10.RelativeRMSEforhourlyresolveddayaheadDNIforecastsbasedon2daypersistence(upper)andECMWFIFS(lowerplanel)in2011.

Page 134: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page134of148

Therefore,withintheMACC-IIprojectanassessmentofthecurrentstatus isperformed.Afocus islaid on direct normal irradiance forecasts being needed by concentrating solar technologies andpoorlyevaluatedsofar.ThisstudyhasbeenextendedwithinMACC-IIItocoveradditionalstationsinNordafrikaandtocoveramorerecentlymadeavailable,denseDNInetworkinSpain(Figure13.10).Apaperhasbeensubmittedforpeer-review.

13.10UVserviceextensionandvalidation

TocomputethesurfaceirradianceintheUltraViolet(UV)spectralrangewithintheIFS,adedicatedradiative transfer codeworks aspost-processorusingas input theMACC forecast fields, includingprognosticOzone,aerosolopticalthickness,cloudfractionandcloudmixingratio.Thecodeprovidesthedown-wellingspectralradiationforthelowestmodellevelintherange280nm-400nmatthreespectralresolutions(0.2nm,1.0nm,5.0nm).ThecurrentoperationalUVprocessorwasdevelopedasamodifiedversionoftheECMWFshort-wave(SW)radiationschemeinusebeforetheIFScycle32R2.

Evaluationof theoperationalUVproductsagainstgroundmeasurementshighlightedbiases in thequalityoftheUVindex(Arolaetal.2014MACCD_122.7):

• The UVB, in particular the range 305-310 nm, is overestimated at noon in clear skycondition (up to ~30% of the observed value). This leads to an high occurrence ofoverestimatedUVindex(whichdependsmostlyonthosewavelengths)

• the quality of the spectral irradiance computation shows a strong dependency on thesolarzenithangle(SZA)withalargeunderestimationforlargeSZA

We investigatedthecauseof theseerrorswithasetofstand-alonesimulationscomparingtheUVprocessoragainstabenchmarkradiativetransfercode.Thesebiasesarelinkedtolimitationsinthecurrentclear-skyRTsolver.TherevisedUVprocessorusesadifferentRTschemefortheclearandcloudyfractionsoftheatmosphericcolumnanditisnowconsistentwiththeRTsolverusedwithinthe operational IFS radiation scheme. Single column experiments show that the revised UVprocessor agrees better than the current UV processorwhen compared to a benchmark RT code(UVSPEC)inclearsky.

Comparison of the spectral irradiance against EUVDBmeasurements in Thessaloniki for the year2004 shows that the new scheme reduces the SZA dependence of the bias and also theoverestimation around 300nm for small SZA (Figure 13.11).On the other hand, the new schemeshowsthetendencytounderestimatetheirradiancebetween20°and40°SZAandoverestimateitfor larger SZA, especially in theUVA region. The latter effect is due to the limitationsof the two-streamapproximation, as shown in ForsterandShine (1995). Similar comparisonsagainstEUVDBmeasurementsinJokioinenandSodankyläshowageneralimprovementwithrespecttothecurrentoperationalcode,withbiaseslimitedto+-20%andconsistentbehaviouracrossdifferentSZA.

Page 135: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page135of148

Figure13.11RatiobetweenEUVDBBrewerandECMWFUVirradianceinThessalonikifortheyear2004,forallskysituations.Currentoperationalcode(lower)andrevisedversion(upper).

Further complete evaluation of the impact of the revised UV processor against surfacemeasurementswillquantifythequalityofthenewUVproductsintheoperationalconfiguration.

280 300 320 340 360 380 4000

0.2

0.4

0.6

0.8

1

1.2

1.4

Wavelength nm

Irrad

ianc

e ra

tio

EC vs EUVDB spectral UV ratio, Thessaloniki, 2004

10−20 SZA20−3030−4040−5050−6060−7070−8080−90

280 300 320 340 360 380 4000

0.2

0.4

0.6

0.8

1

1.2

1.4

Wavelength nm

Irrad

ianc

e ra

tio

EC vs EUVDB spectral UV ratio, Thessaloniki, 2004

10−20 SZA20−3030−4040−5050−6060−7070−8080−90

Page 136: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page136of148

13.11UVservicevalidationproceduresassupporttoVALsubproject

DuringMACC-IIandnowMACC-IIIwehavebeenparticipatingalsoinvalidationactivitiesofVALsub-project, using broadband UV data. Overall, 9 European sites were included in a comparison ofMACC_osuite version (fnyp and from 2014 September g4e2) with respect to COST UV IndexDatabaseandsofarcoveringtheperiodofJanuary2013–December2014.

13.12References

ArolaA.,PitkänenM.,CesnulyteV.,LindforsA.V.,BozzoA.,FinalreportonQC/validationproceduresfortheUVchain,MACC-IIDeliverableD_122.7,2014

Forster P.M. and Shine K.P., A comparison of two radiation schemes for calculating ultravioletradiation,Q.J.R.Meteorol.Soc.,121,1113-1131,1995

Klüser,L.,N.KilliusandG.Gesell,2015,APOLLO_NG-Aprobabilistic interpretationoftheAPOLLOlegacy for AVHRR heritage channels, Atmos. Meas. Tech. Discuss., 8, 4413-4449,doi:10.5194/amtd-8-4413-2015,2015

Oumbe, A., Qu, Z., Blanc, P., Lefèvre, M., Wald, L., and S. Cros, Decoupling the effects of clearatmosphereandcloudstosimplifycalculationsofthebroadbandsolarirradianceatgroundlevel,Geosci.ModelDev.,7,1661-1669,doi:10.5194/gmd-7-1661-2014,2014.

Oumbe, A., Qu, Z., Blanc, P., Lefèvre, M., Wald, L. and S. Cros, Corrigendum to "Decoupling theeffects of clear atmosphere and clouds to simplify calculations of the broadband solarirradiance at ground level" published in Geosci.Model Dev., 7, 1661–1669, 2014, Geosci.ModelDev.,7,2409-2409,doi:10.5194/gmd-7-2409-2014,2014.

Page 137: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page137of148

14.Regionalpolicysupport(POL)

DuringthetransitiontowardstheCopernicusAtmosphereMonitoringService, thedevelopmentoftoolsandprovisionofresultshavetooccuraccordingtotheneedsandrequirementsofusersandpotential users. Focussing on policy users in particular, the objectives of the sup-project POL inMACC-IIIweretofurtherdevelopproductsthatareofrelevancetoEuropeanairqualitylegislation,andtoestablishlinksbetweenproductdevelopers,dataprovidersandusers.

The interactionbetweenPOLand itsuserswas two-way, includinguser feedback tobe taken intoaccount in the development of our products. The provision and development of policy-relevantproductsalsorequiresaconsiderableshareofresearchandtechnicalwork.

ThesectionsofthischapterreflectthemainactivitieswithinPOL,i.e.thetechnicaldevelopmentandmaintenance of services, research activities, and the interaction with existing users as well aspotentialusers.

Linkstopolicy-relevantproductsalongwithfurtherdocumentationcanbefoundatthePolicyPortalathttp://macc.copernicus-atmosphere.eu/services/aqac/policy_interface/.

14.1TheproductlinesofPOL

ThemaingoalsofPOLregardingproductdevelopmentweretoprovideinformationon:

• the efficiency of emission reduction measures to mitigate or to prevent air pollutionepisodesontheshortterm;

• thesourcescontributingtoairpollutionepisodesinthepast,andtopredictedair-pollutioninthenearfuture.

WithinPOLtwoproductshavebeenfurtherdevelopedtoaddressthesegoals:theGreenscenariostoolboxandtheSource-receptorproduct.

14.1.1The‘GreenScenarios’toolbox

The toolbox for emergency control scenarios, or “green scenarios”, is conceived to increaseawareness of policymakers and the general public on thepotential impact of emission reductionstrategiesthatcanlimittheintensityofanairpollutionepisode,ifthesemeasuresaredecidedafewdaysbeforetheepisodeoccurs.ThetoolboxhasbeendevelopedbyINERISandprovides,onadailybasis,Europeanmapspresentingtheexpectedeffect thatshort-termmeasuresonemissions fromvarious activity sectors may have on predicted air pollution. Maps of concentrations of ozone,nitrogen dioxide, as well as particulate matter (PM) corresponding to various assumptions onemission reduction, are available. The simulations are performed as one-, two-, and three-dayforecasts.

If measures are decided sufficiently in advance (several days before the episode occurs) andimplementedover relevantgeographical areas, they canprevent situationswhere limit valuesareexceeded.Inthisrespectitisalsoimportanttodistinguishcontrolmeasuresbyemissionsector.Forexample, it is well-known that PM episodes in spring are due in most cases to high ammoniumnitrate concentrationswhich result fromboth agricultural (ammonia) and traffic (NOx) emissions.

Page 138: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page138of148

Thus itwould bemore efficient to target those sectors during these episodes than the industrialsector.

Currentlythetoolboxconsidersfourcontrolscenarios,assumingemissionreductionsforroadtraffic,residential heating, agriculture, and industrial activities, respectively. The model calculations aredonewiththeCHIMEREairqualitymodel(which ispartoftheMACC-III regionalmodelensemble)usingmeteorological conditions from theECMWF-IFSmodel,boundary conditions from theglobalMACC-IIIproductions,andemissionscenariosbasedon theMACC-IIIEuropeanemission inventoryprovidedby TNO. The emission sectors are reducedby a fixed amount throughout Europe.As anexample,Figure14.1showsfor14July2015thechangeinozoneandPMthatwouldhaveresultedfroma30% reduction in traffic emissions, twodays after implementationof thesemeasures. ThetoolboxisaccessiblefromthePolicyPortalat:

http://macc.copernicus-atmosphere.eu/services/aqac/policy_interface/green_scenarios/

Figure14.1.Changeindailymaximumozone(left)andPM2.5(right)thatwouldresultfroma30%reductionintrafficemissions,twodaysafterimplementationofthesemeasures(16July2015).Unit:μg/m3.

14.1.2Source-receptorcalculations

Importantquestions concerningAirQualitypolicy include i)what causesairpollutionepisodes; ii)cantheybepreventedby localmeasures,andiii)howmuchofthepollutionis importedandfromwhere?

Thesource-receptorproducts,availablefromthePolicyPortalat

http://macc.copernicus-atmosphere.eu/services/aqac/policy_interface/country_sr/

and

http://macc.copernicus-atmosphere.eu/services/aqac/policy_interface/regional_sr/

help answering these questions, either to support short-term action or to give hints on futurelegislationthatcanhelppreventsuchepisodesfromre-occurring.ThemodelcalculationsonwhichtheproductisbasedareperformedbyMETNorwaywiththeEMEPMSC-Wairqualitymodel(which

Page 139: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page139of148

ispartof theMACC-III regionalmodelensemble)using forecastmeteorologyfromtheECMWF-IFSmodelandemissiondatafromtheMACC-IIIEuropeanemissioninventoryprovidedbyTNO.

The two types of calculations quantify country-to-country and regional-to-agglomerationcontributions, respectively. The effects of the main contributing countries (usually neighbouringand/orlargecountries)toforecastedairpollutionepisodesinacountrycanbevisualized,whileforanagglomeration,thecontributionsfromindigenousandimportedairpollutioncanbecalculated.Inbothcases, thecountryoragglomerationcanbechosenby theuser.DuringMACC-III,drop-downmenus for agglomerations have been added and the number of agglomerations, for which thecalculationisdoneweeklyhasincreasedfromtwotosix,whereasintheweborderform,theusercan request other agglomerations as well. The country-scale calculation is currently available ondemandfor48countriesandregions.

As this type of calculation is computationally expensive, the purpose of the calculation has to bespecifiedintheonlineweborderform.

As an example, Figure 14.2 shows results for a summer situation in the Rhine-Ruhr area. Localemission reductions in this case would have led to increases in ozone (due to reduced ozone-titrationeffects),whilereductionsofemissionsoutsidewouldhaveledtoreductions(reductionsoflong-range transported ozone being the dominating effect). These results are strongly dependentnotonlyontheemissionpatternbutalsoontheseasonandmeteorology,andwillthusvaryfromdaytoday.

Figure14.2.Effectofa15%reductionofemissionswithintheRhine-Ruhrarea(left)andoutsidetheRhine-Ruhrarea(right)onozonelevelswithintheRhine-RuhrareainJune/July2015.Changesaregiveninpercent,averagedovertheRhine-Ruhrarea.

During MACC-III, a new capability has been developed to visualize the effect of long-rangetransported pollution as animations. These animations assist the user in the interpretation of thetypeofbarchartsshowninFigure14.2,astheycanshowmoreclearlywhereairpollutionoriginatesfrom for a given meteorological situation. Following a user request, sources outside theagglomeration are further subdivided into national and international sources. An example filetogetherwithdocumentationhasbeenaddedtothePolicyPortal.

14.2ResearchinPOL

POLalso includedelementsofresearch,whicharerelevanttoAirquality legislation. Forexample,whendesigningemissionreductionprotocols it is importanttohaveagoodunderstandingoftheirefficiency in reducing air pollution. The effects of fictive emission reductionmeasures have beenstudiedbyCNRS-LISAfortwoairpollutionepisodesthatoccurredinthepast,andasummaryreport

Page 140: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page140of148

hasbeenputonthePolicyPortal.Thestudyconcludesthatforbothepisodes,onecriticalsectorofactivitytoconsiderforemissioncontrolinthesesituationswouldhavebeenagriculture,whiletrafficandresidentialheatingwouldhavebeenlessefficient inreducingairpollution. Inthesesituations,airmassstagnationwasnotapredominantmeteorologicalfeature.AsfortheGothenburgprotocol,thisscenarioshowsencouragingresultsforthereductionofairpollution,onceallthecountrieshaveratified the protocol. The highest effect of emission reductions on regional air composition issimulatedafewdaysaftertheimplementationofemergencyemissioncontrolmeasures.Ingeneralterms,exceptforregionallydiffuseemitters,reducingemissionsoncetheepisodeisdevelopedmaybe inefficient. By contrast, when the decision is taken before the episode occurs, significantreductions inPM10arepossible for thisepisode.Theresults thus indicate that the largestbenefitsareobtainedwithlong-termactionsonemissions.

14.3InteractionwithAirQualitypolicyusers

The development of products relevant to policy depends strongly on feedback from policy usersregardingthetoolsbeingdevelopedandtheresultsthatareprovided.Furthermore,foraservicetobesuccessfulitisimportanttodemonstratetheavailabletoolsandtopromotetheiruse.

The interaction with policy users was thus an important element of POL. Main activities duringMACC-IIIweretheimprovementofthePolicyPortalandtheorganizationofasecondworkshopforpolicyusers(andpotentialusers)withintheEuropeanAirQualitypolicyinfrastructure.

14.3.1ThePolicyUserportal

Sinceearly2012thePolicyPortalcanbeaccessedat:

http://macc.copernicus-atmosphere.eu/services/aqac/policy_interface/

DuringMACC-III,METNorwayfurther improvedthiswebpagetosimplifyaccesstopolicy-relevantproducts not only from POL but also from other subprojects of MACC-III (e.g. chemical weatherforecasts,annualAirQualityassessments,alertservices). Inaddition, informationonPOLactivitiesthatarerelevantforpolicyusersispresented(e.g.reportsfromthePolicyUserworkshops).

14.3.2ThePolicyUserworkshopofMACC-III

ThePolicyUserworkshopofMACC-IIItookplaceon03/04March2015.ItwasorganizedbythePOLpartnersNILU,EAA,METNorwayandINERISandhostedbyEAAinVienna.

It attracted 29 product providers, users and potential users from 13 European countries. Theworkshop focusedondownstreamuserswhomakeuseofMACC-III data forpolicy support.MostparticipantshadalreadygainedsomeexperienceinusingMACC-IIIproducts,suchastheTNO-MACCemission data, the MACC global model or regional ensemble output for chemical boundaryconditions,andtheairpollutionforecasts(inparticularduringepisodes).

During the first session of the workshop, possibilities of collaboration betweenMACC-III and theFAIRMODEcommunityaswellasanumberofusecaseswerepresented,whilethesecondsessionpresented news about policy-relevant products of MACC-III, such as chemical weather forecasts(includingpollen),greenscenarios,emissionsandsource-receptorproducts.Theseconddayoftheworkshopwasdedicatedentirelytouser-providerinteraction.Duringaseriesoftrainingsessionsthe

Page 141: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page141of148

audiencereceivedamoredetailedintroductionaboutwheretofindMACC-IIIproductsandhowtouseandinterpretthedataandgraphs.Thedifferentroundtablediscussionsofthefinalsessiongaveall participants the opportunity to provide their feedback and to make suggestions for MACC-IIIproductrefinementsaswellasnewproducts.DuringandaftertheworkshoptheproductprovidersofMACC-IIIreceivedveryusefulcommentsfromrepresentativesfromninedifferentcountries,usersthatwillbeabletosetthepathfortheinformation,whichtheirnationalpolicyofficerswillrequire.Among the main requests from the users were: the timeliness of the products, the tailoring ofproductstospecificregionsandtosimplifyaccesstoproducts.Moredetailedinformation(includingtheagenda,allworkshoppresentations,andaworkshopreport)canbefoundatthePolicyPortal:

http://macc.copernicus-atmosphere.eu/services/aqac/policy_interface/second_pol_workshop/

InadditiontothePolicyUserworkshop,thepolicy-relevantproductsofMACC-IIIwerepresentedattheACCENTplusResearch-PolicyStakeholderMeetinginBrusselson2December2014.MACC-IIIwasoneof5Europeanprojectsrepresentedattheworkshop.TheagendaandthepresentationsgivencanbefoundattheACCENTpluswebsite.

14.4Summaryandoutlook

Overall, POLhas achieved its objectives. Thepolicy-relevant products ofMACChavebeen furtherdeveloped and demonstrated during MACC-III, links between providers and potential users havebeenestablishedand/ormaintained,and theusefulnessofMACC-IIIproductshasbeenconfirmedby the European air quality policy arena. Important feedback on available productswas receivedduring MACC-III, which also will support the further development of policy-relevant products inCAMS.

Future challenges will be to find a balance between the needs and possibilities within the givenresources.RegardingtheproductsthathavebeendevelopedinPOL,themainlimitingfactoristhelargecomputationalrequirementratherthanalackofscientificknowledgeortechnicalfeasibility.

Apartfromthetechnicaldevelopmentofproducts,theinteractionwithusersandtheidentificationofmoreuserswillbeofparamount importance inCAMS.During theCAMSprecursorprojects theusefulness of these products for air quality legislation has been confirmed numerous times, andCAMSwillprovideanexcellentopportunitytofurtherpromotetheiruse.

Page 142: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page142of148

Page 143: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page143of148

15.UserInterfaceactivities(INT)

The“UserInterface”sub-projectofMACC-III isessentialandcentraltotheentirelinkwithusersinMACC-III including understanding user requirements, coordinating user involvement, organizing atechnical user interface, coordinating project communications and training activities, so that theentireMACC-IIIprojectreachesthebestpossibleuseofitsproductsbyitsusers.TheuserinteractioninMACC-IIIcontinuedwiththehybridapproachbetweenbestpractices inthesatellitecommunity(basedonsystematicanalysis+documentationofuserrequirementsandsystemspecifications)andthe NWP community (based on continuous monitoring of skill scores and communication withusers). This reflects the need for continuous iteration and improvement of the understanding onproviderandusersideandassuresaregularadvancementofbothcapabilities.

15.1MACC-IIIUsers

UsersofMACC-IIIservicescomefromawiderangeofcountries,sectorsandtypesoforganizations.MACC-IIIusersvaryfrom‘power’users(requiringdailydata)touserslookingforspecificdata(area,time,species).Dataaccessisfullyopenandfree.ThefollowingtableprovidesanoverviewofMACC-IIIusers–basedonuser registration (name, institute/company,country) implemented foralldataservers. Table 15.1 provides numbers of users (or of requests) for MACC-III main categories ofproducts.

Table15.1.SelectedMACC-IIIfiguresonusersandusage(attheendofMACC-III,June2015).

Service NumberofUsers/RequestsfordataGlobalNRTAnalyses&Forecasts ~225usersRegionalNRTAnalyses&Forecasts 155usersGlobalReanalysis 1600usersGHGfluxinversions 40usersSolarRadiation ~1000requests/yearGlobalftp ~40usersEmissions,fire 1773users(716institutes)Userqueriestrackingsystem 436itemsuntil22/6/2015

ThegeographicaldistributionofMACCusersisshowninthefollowingtwomapsonglobalandEuropeanlevel(Figure15.1).

Figure15.1.Geographicaldistributionsofusersattheglobalscale(left)andinEurope(right).

Page 144: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page144of148

MACC-IIIuserscangenerallybesplitincommercial,governmental,science,andother;theirdistributionoverthemajorMACC-IIservicesisshowninFigure15.2.

Figure15.2.SplitofMACC-IIIusersperhigh-levelcategoriesfortheglobal,regionalandsolarradiationproducts.

In MACC-III the full set of user documentation was integrated / updated: a comprehensive userrequirementsdocument,auserfeedbackdocument,auserqueriessummary,ausermeetingreportandtwoUserAdvisoryBoardteleconreports.TheMACC-IIIuserworkshopwasheldinItalyinMay2015,co-organizedbyISPRAwhoiscoordinatingenvironmentalagenciesandresearchinstitutesinthecountryontheuseofCopernicusservices.Theuserworkshop,togetherwithanotherdedicatedpolicyuserworkshoporganizedby thePOLsub-project inVienna inMarch2015broughtvaluableuser feedback (enhancedunderstandingofuserneeds, confirmedusefulnessofMACC-IIIproductsand services, needs for evolution). Two telecons of the MACC-III User Advisory Board wereconducted which led to helpful recommendations. All together the user requirements documentcontains now 288 numbered requirements: 188 technical, 44 regarding data access, 6 concernedwithdatapolicyand50forthenewlyaddedcategoryuserinvolvement.Theuserqueriesanalysisinthejirasystemcurrentlyholds436userqueries.TheevolutionofquerytopicsisdepictedinFigure15.3showingthepercentageofdifferenttopics(leftupto7/2014,rightupto6/2015):Itisobvious,thatdatarequestscontinuetobethemajorityofrequests,whereasthenumberofqueriestargetedattechnicalissues,errorsandquestionsincreasedfrom15%to25%.

Figure15.3.TopicsofuserqueriesregisteredintheJirasystemuptoJuly2014(left)anduptoJune2015(right).

Page 145: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page145of148

The user feedback collected at the different levels of user interaction fromdirect contacts of theservice leaders togeneraluserworkshopsand feedbackwasanalysed;due to the shortperiodofMACC-IIInomajorserviceevolutionwasconducted;majoruserfeedbackwashoweverdocumentedtobe further assessed inCAMS.AsdecidedduringMACC-II, a simpleonlineuser registrationwasimplemented toall technical interfaces (shortuser identificationandacknowledgementofgenerallicenseterms),withnoformalServiceLevelAgreement(SLA)necessaryforgeneraluseofstandardoutput.SpecificSLAsremainimportanttodocumentdedicateddeliveryagreedwithselectedpowerusers;thismechanismwillmainlybeofuseinthecomingoperationalphase(sofaronlyafewSLAsor similar agreements have therefore been concluded). At the end ofMACC-III all user interfacedocuments(requirementscollection,userqueries,userfeedbacksummary)havebeenbroughttobeinclusive of all input receivedduringMACC-III. An example use case documentedduringMACC-IIIuserfeedbacksummaryreportisshownbelow(Figure15.4).

Figure15.4.ExampleofusecasesheetdevelopedduringMACC-III.

15.2Websiteandtechnicalinterfaces

The technical interfaces to MACC-III output were further advanced: metadata were furtherharmonized between differentMACC-III service lines and dissemination interfaces aswell aswithINSPIRE/GEO/WMO-WIS. Inresponsetouserrequests thewebportalcontainsoneharmonizedsearchablecatalogueforallproductslinkingtothedifferenttechnicalinterfaces,usingthestandardmetadataandalso linking to furtherdocumentationon thewebportal.Standardcompliant (OGC)inter-operable metadata interfaces and web services for the MACC-III products have beenmaintained:theintegratedMACC-IIIcatalogueatECMWF,theglobalboundaryconditionsserveratFZJ,theEuropeaninterface(ensembleandvalidatedassessments)atDLR’sWDC-RSATbasedonthe“interfacetocore”ofthePASODOBLEdownstreamproject,solarirradianceserverMcClearatEcole

Page 146: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page146of148

deMinestoextractlocalirradiancetimeseries.TwoteleconswereheldbetweenrepresentativesofthetechnicaluserinterfacesofMACC-IIIservicestodiscusstheremainingmetadataissuesandtheirharmonization. In thegraphical interface (MapViewer)ausercannowdefinea rectangular spatialsubset. After choosing a data format, the “Download-Wizard” gives the user the opportunity toeither select the full domain or the spatial subset (Figure 15.5). Moreover the user may set thespatial resolution according to his requirements. Moreover, access to EVA products as providedworking group through the UI is operational. Modifications of the spatial extent of the domains(2007…2010vs.2011)arehandledbythevieweraswellasbythedownloadwizard.AttheendofJune201542,usersareregisteredforMACC-IIIattheWDC-RSATuserinterfacemaintainedatDLR.Extended promotion of the capabilities of this user interface is recommended, now that itimplementsmost important user requirements (e.g. data download of subsets determined in thegraphicalUI).

Figure15.5.Download-Wizardforspatialsub-settingfromtheGUIatDLR/WDC-RSAT.

Themost significant update to the web portal was provided by ENS with entirely new access tographics.TheimplementationofthenewregionalwebsiteintothemainMACC-IIIwebsiterequiredcarefulcoordinationtoensurealllinks(bothonthewebpagesandintheCatalogue)werecorrectlyupdated or established. Another new item is the provision of access to data and plots for theEuropeanemissiondataset.MACC-IIIhascontinuedtopromotetheprojectthroughawiderangeofmediaandinteractions.Amongothers,MACC-IIIprovidednewsstoriesandotherinformationonitswebsite,newsletters,contributionstoothernewsletters,contributionstotheEuropeanSpaceExpo,and contributions to various internationalmeetings.MACC-IIIwas representedatawide rangeofinternationalmeetingstargetingdifferentusercommunities.MACC-IIIhascontinuedtousethewebsiteathttp://www.copernicus-atmosphere.euasitsmaininterfacetousers.Thesiteprovidesaccesstodataandgraphicsforalltheservicesaswellasincreaseddocumentation.Partofthewebsiteisdedicated to routine users providing information on system upgrades, validation and verification,andoperationaldataaccess.StatisticsofvisitstothewebsiteareshowninFigure15.6for2014.

Page 147: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page147of148

Figure15.6.VisitsstatisticstotheMACC-IIIwebsitein2014.

15.3Communicationsandtraining

TwoissuesoftheMACCNewsletter,MACCazinewerepublished.OneappearedinOctober2014andoneinJanuary2015.Theyaresenttoabout1500recipientsandhaveanapproximateOpenrateof32%andaClick rateof6%.Botharewell above theaverage rates fornewsletters like these.TheJanuaryMACCazineisshowninFigure15.7.

Figure15.7.TheJanuaryissueoftheMACCazine.

Page 148: Final Report MACC-III

MACC-IIIFinalReport(February2016)

Page148of148

Focusedtrainingbasedonuser requestwasprovidedat theMACC-IIIPolicyUserworkshopwhichtook place in Vienna on 3-4March 2015 and gathered 29 product providers, users and potentialusersfrom13Europeancountries.Thetrainingsessionsorganizedaspartofthisworkshopincludeda short overview of the MACC-III website and its elements (http://copernicus-atmosphere.eu), adetailedaccountofhowtouseandinterprettheannualairqualityassessmentreports,thedailyairquality forecasts (includingthenewuser interfacewhich iscurrentlyunderdevelopment),andthesourcereceptorproductofMACC.Thetrainingsessionsweregivenaslivepresentations,usingtheweb pages themselves via the wireless network of the meeting venue. Screenshots of thepresentations have not beenmade. During the training sessions, participantswere invited to askquestions, toenablean interactivediscussion.Someselectedslidesareavailableontheworkshopweb page. As part of the ECMWF training courses, specific lectures on modelling and dataassimilationforatmosphericcompositionweregiven.Thelistofthedifferentcoursesisavailableat:http://www.ecmwf.int/en/learning/training. These training courses are targeted on early-careerscientistandareopentoparticipantsacrosstheworld.SeveraldiscussionstookplaceduringMACC-IIIonthedevelopmentofaFAQ(FrequentlyAskedQuestions)sectionontheMACC-IIIwebsite.Inafirst stepaconceptwasdefinedwhich included themotivation fordesigningaFAQ,howtheFAQcould be designed, the topics that could be covered by the FAQ togetherwith details on each ofthese topics, and a few ideas on how to present and start the FAQ pages. This concept waspresentedtofewselectedusersandtheirfeedbackwascollected.Finally,aprototypewasdefinedfortheexampleoftheemissionproductsandservice.