学术会议海报模板4academic conference poster model

PrototypingSSTRetrievalsfromGOES-RABIwithMSGSEVIRIDataNikolayV.Shabanov1,2,Nikolay.Shabanov@noaa.gov,(301)763-8102×154AlexanderIgnatov1,BorisPetrenko1,2,YuriKihai1,3,XingMingLiang1,4,WeiGuo1,2,FengXu1,4,PrasanjitDash1,4,MichaelGoldberg1,JohnSapper51NOAA/NESDIS/STAR;2IMSGInc;3PerotSystemsGovernmentServices;4ColoradoStateUniversity-CIRA;5NOAA/NESDIS/OSDPDGeostationaryOperationalEnvironmentalSatellite-RSeries(GOES-R)willcarryAdvancedBaselineImager(ABI)onboard.SeaSurfaceTemperature(SST)algorithmforABIisbeingdevelopedbytheSSTTeamwhichisapartoftheGOES-RAlgorithmWorkingGroup(AWG).ABISSTproductionisprototypedwiththeMeteosatSecondGeneration(MSG)SpinningEnhancedVisibleandIRImager(SEVIRI)data(Schmetzetal.,2002).TheAdvancedClear-SkyProcessorforOceans(ACSPO)developedatNOAA/NESDISandcurrentlyoperationalwithAVHRRdataonboardNOAA-18andMetOp-AhasbeenadoptedtoSEVIRI.ACSPO-SEVIRIsystemprocesses15-minutefull-disk(FD)SEVIRILevel1datainnear-realtime(NRT)andgeneratesLevel2clear-skyproductsoverocean,includingtop-of-atmosphere(TOA)clear-skybrightnesstemperatures(BTs),SSTandaerosols.ThisposterevaluatesinitialBTandSSTretrievalsfromMeteosat-9SEVIRIforonefullmonthofdatainJune2008.Theviews,opinionsandfindingscontainedinthisreportarethoseoftheauthorsandshouldnotbeconstruedasanofficialNOAAorU.S.Governmentposition,policy,ordecision.ThisposterdoesnotreflecttheviewsorpoliciesoftheGOES-RProgramOfficeorAlgorithmWorkingGroup.ACSPO-SEVIRISystemCRTMsimulationsforSEVIRIlongwavebands9(11μm)and10(12μm)areconsistentwithsimilarresultsfortheAVHRR.Moreanalysesareneededtobetterquantifythesepreliminaryobservations.M-ObiasinCh4(3.7μm)isinconsistentwithAVHRRandrequiresfurtheranalyses.ComparisonwithGlobalReferenceSSTsTheSSTQualityMonitor(SQUAM)currentlyroutinelygeneratesconsistencystatisticsbetweentheretrievedAVHRRSSTandmultiplereferenceSSTfields(Dashetal.,2009).SQUAMwasappliedtoquicklyevaluateSEVIRISSTretrievals.ThefollowingreferenceSSTwereused:DailyReynolds:OISST(AVHRR-based)andOISST-A(AVHRR+AMSR-E),OSTIA,RTGHR(HighResolution)andLowResolution(LR),andPathfinderSSTclimatology.SSTDiurnalCycleIntroductionCRTMandBTSimulationsFig.2.ExampleofMSG-2SEVIRI15-minFDdataforJune14,2008,2:00pmUMT,processedbytheACSPO-SEVIRIsystem.(a)Visibletrue-colorimageconstructedfromalbedo.BlackstripexpandingfromSEcorrespondstonight-time;(b)CloudMask;(c)SST;(d)AODretrievedfromCh01usingsingle-channelalgorithm(DePaepeetal.,2008).Dashetal.,(2009).TheSSTQualityMonitor(SQUAM),submitted,RSE.DePaepeetal.,(2008).AerosolretrievaloverOceanfromSEVIRIfortheuseinGERBEarth’sradiationbudgetanalysis,RSE,112,2455-2468.Liangetal.,(2009).ImplementationofCRTMinACSPOandvalidationagainstnighttimeAVHRRradiances,submitted,JGR.Merchantetal.,(2008).Optimalestimationofseasurfacetemperaturefromsplit-windowobservations,inpress,RSE.Schmetzetal.,(2002).AnIntroductiontoMeteosatSecondGeneration,BAMS,83,977-992.Waltonetal.(1998).ThedevelopmentandoperationalapplicationofnonlinearalgorithmsforthemeasurementofseasurfacetemperatureswiththeNOAApolar-orbitingenvironmentalsatellites.JGR,vol.103(C12),27999-28012.DisclaimerLiterature(a)(b)(d)(c)InputtoACSPO-SEVIRIsystemis15-minSEVIRIFDchanneldata(opticalCh1-3andthermalCh4,9-10).1°-resolutionweeklyReynoldsSSTand6-hourNationalCentersforEnvironmentalPredictionGlobalForecastSystem(NCEP/GFS)dataareusedasinputtothefastCommunityRadiativeTransferModel(CRTM)tosimulateclear-skyBTinCh4,9-10(Liangetal.,2009).TwoSSTalgorithmsareimplemented.TheRegression(split-window,NLSST)algorithmisbasedonWaltonetal.[1985]equation:Currentlyusedregressioncoefficientsa0–a3werederivedbyEUMETSATforMeteosat-8.ThePhysicalalgorithmusesCRTMtoinvertBT9andBT10forSSTandwater-vaporopticaldepthscalingfactor(Merchantetal.,2008).Theregressionandphysicalalgorithmswillbecross-evaluated,togenerateasuperiorqualityhybridSSTalgorithm.Forthisstudy,asimplifiedcloud-maskwasimplementedwhichcurrentlyusesonlytwotests:SSTtest(comparisonofretrievedandReynoldsSST)andBTtest(comparisonofCRTMsimulatedandmeasuredBTs).Theend-to-endNRTprocessingsystemhasbeensetupfordatastreamprocessing:(1)downloadingSEVIRIdatafromNOAAoperationalservers(incollaborationwithAWGLandTeam),(2)dataprocessingatSSTTeamservers,(3)dataanalysiswithweb-basedQCtools.SEVIRIdatafor2008areonline,2006-2007areonexternaldisks.Theproductfilecontainsupto32datalayers,filesizeofoneFDproductfileis~450MB.Executiontimeis~7minperFDonDELLPowerEdge2900.Fig.6.SSTbiasasfunctionofNumberofAmbientClear-Skypixels(NAC;calculatedwithin21×21GACpixels).FDdatafrom1-7June2008arebinnedatNAC=20andaveragedfor2hourscenteredonMiddayandMidnight.Fig.3.SolidlinesshowFDmonthlymeandiurnalcyclesof“RetrievedminusReynoldsSST”.IndividualdatapointscorrespondtodifferentdaysinJune2008.SST=a0+a1BT9+a2TReynolds(BT9-BT10)+a3(BT9-BT10)(sec-1),FirstGuessSSTReferenceSSTSEVIRIDynamicData:Opticalbands[1,2,3]Thermalbands[4,9,10]Illuminationgeometry(SZA,RAA)(HDF,5x5km(atnadir),15min)SEVIRIStaticData:Land/WatermaskViewgeometry(VZA)Lat/Lon(HDF,5x5km(atnadir))AncillaryGFSData:AirandSurfaceTemperatureWatervaporWindspeedOzoneconcentration(HDF,1°x1°,6-hours)Regression(Splitwindow)Physical(CRTM-based)CRTMsimulationsofBTandBTJacobianatcoarse(1°x1°)GridCloudMaskAlgorithm(7tests)1)SST(currentlyon)2)BT(on)3)Ch3Albedo(currentlyoff)4)ReflectanceRatio(off)5)Three-Five(off)6)UltraLowStratus(off)7)Uniformity(off)ChannelCalibrationBi-LinearInterpolation(BTandBTJacobian)1ox1o->5x5km(atnadir)SingleChannel6S-basedLUTsOutput(upto32layers/SDS):TOAalbedoinVISbands(1,2,3)TOABTatThermalbands(4,9,10)SST(Regression&Physical)AerosolOpticalDepthsinVISbands(1,2,3)CRTBTsGeometryandGeolocationQC(cloudmask,L/W,day/nightflag,etc)OISST(Reynolds)GFSfieldsinterpolatedtoSEVIRI(HDF,5x5km(atnadir),15min)AncillaryOISSTData:ReynoldsSST(HDF,1ox1o,weeklyor0.25ox0.25o,daily)SSTAlgorithmAODAlgorithmWeb-basedQCTools:ErrorcharacterizationasfunctionofEnvironmentalConditions(SSES)SSTevaluationwithreferenceFieldsandBuoyData(SQUAM)BTevaluationwithCRTMsimulations(MICROS)Fig.1.Flow-chartoftheACSPO-SEVIRIsystem.ReferenceSSTBi-LinearInterpolation(ReynoldsSST)1ox1o->5x5km(atnadir)IndividualCloudMaskTestsresultsAODflagL/WMaskNay/mightflagGlintflagQCMaskRegressionPhysicalFig.4.TimeseriesofSSTanomaliesevaluatedoverclearskypixelsforJune2008.RetrievedSSTexhibitsthediurnalcyclewithanaverageamplitudeof0.3°C.SSTanomalychangesfromdaytodayandasafunctionoftimeofday.Fig.5.Physical-ReynoldsSSTasafunctionofwindspeed.FDdatafrom1-7June2008arebinnedatV=1m/sandaveragedfor2hourscenteredonMiddayandMidnight.Atnight,SSTisafairlyinsensitivetowindspeed.Duringdaytime,SSTchangesasafunctionofwindspeedwithanamplitudeof~0.7K.Differencebetweendayandnightislargestatlowwinds,duetoformingastrongdiurnalthermocline.Thehigherthewindspeed,thesmallerthediurnalheating.SSTdependsuponambientclear-skyconditionsduringbothdayandnight(Xueta.,2009).Thediurnalsignalissmallestwhenambientcloudinessishigh(NAC→0)andlargestunderclearskies(NAC~400).JP2.1589thAMSAnnualMeetingand5thGOESUsers'Conference,Phoenix,AZ,11-15January,2009Correspondence:Nikolay.Shabanov@noaa.gov,Tel.:301-763-8102x154,Fax:301-763-8572Fig.8.TimeSeriesofM-ObiasesinSEVIRICh4,9and10inJune2008.(NighttimedatawithVZA<70°).Therootcausesofthetwodrop-outanomalieson11and15June2008arebeinginvestigated.CommunityRadiativeTransferModel(CRTM)isusedinACSPOinconjunctionwithReynoldsSSTandNCEPGlobalForecastSystem(GFS)upperairfieldstosimulateclear-skyBTsinSEVIRICh4,9,and10.CRTMBTsareusedforcloudmasking,physicalSSTretrievals,andtomonitorqualityofSEVIRIradiances.Here,Model(CRTM)minusObservation(SEVIRI)(“M-O”)biasesarepreliminarilyevaluatedusingnighttimeSEVIRIdata(SunZenithAngle>90°).Fig.7.Nighttime“M-O”biasesinSEVIRICh4,9,and10foroneFDimagetakenon14June2008@3:00am.DatashownareforSatelliteVZA<70°.HighM-Obiasesof~+0.6KinCh9(11μm)and~+0.3KinCh10(12μm)areconsistentwithAVHRR(Liangetal.,2009).TheymaybeduetoCRTM(missingaerosol,usingbulkratherthanskinSST,notcorrectingSSTfordiurnalcycle)orSEVIRIdata(residualclouds).ThelowbiasinCh4(3.7μm)isinconsistentwithAVHRRanalyses.TheSQUAMmethodologywillberoutinelyappliedtoanalyzeSEVIRISSTs.