Working Together on the Stratosphere:Comparisons of RO and Hyperspectral IR Data in
Temperature and Radiance Space
Michelle Feltz, Robert Knuteson, Johannes Nielsen1, Lori Borg, Thomas August2, Tim Hultberg2, and Antonia Gambacorta3
UW – Madison SSEC / CIMSS,ROMSAF1, EUMETSAT DARMSTADT2, NOAA STAR3
21-27 Sept. 2017, Estes Park, COJoint COSMIC 10th Data Users' Workshop & IROWG-6
• Motivations• Methods• Results
– 1]IRsoundertemperatureretrievalassessment– 2]NASAAIRSandUCARCOSMIC6yr comparison– 3]ROMSAFGRAStemperatureassessment– 4]UCARCOSMICversionassessment
• Conclusions• PathsForward
Roadmap
• Upper-airtemperatureisanEssentialClimateVariableaccordingtotheWMO
• Twotechnologiesofferthepotentialforrecordingclimatequalitystratospheric“benchmark”measurementsasdefinedbytheNRCdecadalsurvey(2007):
1]radiooccultation(RO)2]hyperspectralinfraredsounders(IRsounders**)
• Thisworkcomparesthesetwoindependentmeasurementtechniquesinordertoassesstheirdifferenceswithinthestratosphere
Motivations
**Useof‘IRsounder’inthispresentationreferstohyperspectralIRsounders
• Useprofile-to-profilematchupmethod• AccountsforROprofilegeometryandhorizontalresolution• <1hr timecriterion
FORMOREDETAILS:Feltz,M.etal.(2014),AmethodologyforthevalidationoftemperatureprofilesfromhyperspectralinfraredsoundersusingGPSradiooccultation:ExperiencewithAIRSandCOSMIC,JGR,doi:10.1002/2013JD020853.
Methods:MatchupScheme
Individual Matchup Case
IR sounder fields of view averaged to create profile
GRAS RO profile location
• Methodapplicabletodatafromdifferentplatforms/processingcenters• Distributionandnumberofmatchupsdependsonorbitalmechanics• Previouslyuseddatasets:
RO:UCARCOSMIC,JPLCOSMIC,ROMSAFGRAS,IR:NASAAIRS,EUMETSATIASI,NOAACrIS NUCAPS,…
Methods:MatchupScheme
Matchup Distributions DJF (2015-2016)
Orbital mechanic
differences
GRAS-A / CrIS GRAS-A / IASI-A (thinned)
• UseOSSfastRTM– Radiances&temperatureJacobians
calculatedfor15umCO2 absorptionregion(representsUTLS)
– Inputs:ERA-Interim,NOAACarbonTracker
• TemperatureJacobiansareusedtoassignheightlevelstoradiances
• Methodologicaluncertaintyislargerforchannelswithmoreweightfromabove10hPa duetotheextentoftheROprofileheights– Focusonchannelswithweightbetween
~200–10hPa
Methods:RadianceCalculations
Result1:IRSounderAssessment
FORMOREDETAILS:Feltz,M.L.,etal.(2017),AssessmentofNOAANUCAPSupperairtemperatureprofilesusingCOSMICGPSradiooccultationandARMradiosondes,J.Geophys.Res.Atmos.,122,doi:10.1002/2017JD026504.
IR Sounder – UCAR COSMIC Dry TemperatureDJF Tropics (2012-2014)
Bias (—)Stdev. (- - -)
• ROusedasacommonreferencetocompareoperationalIRretrievals• Averagingkernelsshouldbeappliedtomoveproductstoacommonverticalgridandleadsto
goodagreement(<0.5K)between200-10hPa• ROsuccessfullyusedtoassessIRsoundererrorcharacteristics
Colored by IR instrument:NOAA CrIS NUCAPSNASA AIRS v6EUMETSAT IASI v6 10
100
200
Pres
sure
(hPa
)
0 0.5 1 1.5
Pre
ssur
e (h
Pa)
10-2
10-1
100
101
102
103
Mean Sum of AK rowsIASIAIRSCrISIASI+/-2*StdevAIRS+/-2*StdevCrIS+/-2*Stdev
0 0.05 0.1
10-2
10-1
100
101
102
103
IASI AK
0 0.05 0.1
10-2
10-1
100
101
102
103
CrIS AK
0 0.05 0.1
10-2
10-1
100
101
102
103
AIRS AK
ApplyAveragingKernels
10
100
200
35 hPa: AK*(COSMIC Dry-AIRS) Temp bias666.7 cm-1: AIRS calc – AIRS obs. BT bias
COSMIC calc – AIRS obs. BT biasRobust-Lowess Filtered Time Series
• AIRScalc - obs BTbias• within0.5Kforalllat.zonesà impliesno
largemethodologicalerrors
• COSMICcalc – AIRSobs BTbias• within1Kintropics/mid-lats and2Katpoles• seasonaldependence• similarmagnitudetoAK*(COSMIC-AIRS)temp• à suggestsCOSMICtempisdominating
contributorto35hPa AKsmoothedtempdiff
• IRradianceshave~0.5Ksystematicuncertainty(Tobin,JGR,2014)– furtherinvestigationneededinattributionofseasonaldifferences
Result2:UCARCOSMICv2010vs.AIRSv6(6-years)
Feltz,M.etal.(2017),AssessmentofCOSMICradiooccultationandAIRShyperspectralIRsoundertemperatureproductsinthestratosphereusingobservedradiances,J.Geophys.Res.Atmos.,122,doi:10.1002/2017JD026704.
Result3:ROMSAFGRASvs.IASI,CrIS BTs
• ROMSAFGRASBTscomparedtoIASIandCrIS observations• GRAS-Avs.IASI-A• GRAS-Bvs.IASI-B• GRAS-Avs.CrIS• GRAS-Bvs.CrIS
• Resultsfromfourmatchupdatasetswereconsistent(w/in~0.25K,sharedgeneralcharacteristics)
• TotalBTuncertaintyestimatedtoassessagreementofROandIRBTs• Basedon
• IRstochasticuncertainty• ROstochasticuncertainty• Methodologicaluncertainty
GRAS-B – IASI-B observed BTs(2015-2016)
Result3:ROMSAFGRASvs.IASI,CrIS BTs
Bias (bold markers) Stdev (light markers)Stochastic Unc. (error bars)
• Wetandbackground(bg)BTstatsshowsimilarity• Wetandbg tempdifferencemayhavetoohigh
verticalresolution
• DryBTbias• oftenlargerthanwet/bg bias,reachlargest
valuesat~20hPa• negativeminimainsummersolstices,
positivemaximainwintersolstices
• Wet/bg BTstdev.• typicallyconstantwithheight,roughly0.5K• agreeswithBTunc.inmid-lats/polarwinters• slightlysmallerthanexpectedinpolarsummers
(by~0.1K),slightlylargerintropics(by~0.1K)
• DryBTstdev.• growswithhgt.over150-10hPa• largerthanBTunc.forallzonesforchannelswith
contributionsfromabove70hPaà BTuncertaintyestimatedoesnotaccountforstochasticerrorsourcesofthedrytemperature
JJA DJF
• UCAR COSMIC v2013• UCAR COSMIC v2010*Vertical lines represent span of WF
• Updateimprovesbiasthroughmostseasons,heights,andlat.zoneswithexceptionforTropicswherecoldbiasisincreasedinsomeseasons
• When0.5KusedassystematicuncertaintyforIRsounder,thenconclusionthatUCARv2013drytempisanimprovementuponv2010canbemade
Result4:ROProcessingVersionAssessment
DJFBTBias:- decreasedinArcticbtwn 1-100
hPa byupto0.5K- increasedslightlyinTropicsby
maxof~0.2Kbtwn 10– 1hPa
COSMIC – AIRS observed BT Bias (2011)
Conclusions
• Result1:– UCARCOSMICdrytempusedascommonreferencetoassessIRsounderretrievalerrors.
• Result2:– 6yearcomparisonofUCARCOSMICvs.AIRSobservedBTbiasshowedseasonaldependencyat35
hPa. COSMICBTbiaswassimilarmagnitudeastempdifferenceandreachedoverthe0.5KIRmeasurementunc insomeregionsà furtherinvestigationonattributionofseasonalvariationsisneeded.
• Result3:– ROMSAFGRASBTswerecomparedtoIASIandCRISobservationsusinganestimatedBTuncertainty.– OverallreasonableagreementseenbetweenGRASandIRBTs.– Largestinconsistencyseeninincreaseddrytempstandarddeviationabove70hPaà need
investigation.– Regionsofincreasedbiases,inparticularpolarregionsaroundsolsticesandlowerstratospheric
tropicalchannelsshouldbeinvestigated.
• Result4:– AIRSobservationsusedtoassessUCARCOSMICprocessingversionupdate.COSMICv2013showed
improvedBTbiasagreementwithAIRSoverlargemajorityofseasons,particularlythepoles.
PathsForward
• Investigateareasofconcernandinterestasnotedfrompreviousstudies• e.g.seasonaldependentbiasesinpolarzones,increaseddrytemperaturestandarddeviations,…
• Takeintoaccountchallengesofradiativetransfer,IRandROmeasurementdifficulties
àWhatcanweattributethedifferencesofIRandROto??
• LookatotherIRchannelswhichrepresenttheloweratmosphere
• ContinuedcollaborationbetweenhyperspectralIRsounderandROcommunitiesinhopesforfruitfulinsights
Working Together on the Stratosphere:
Comparisons of RO and Hyperspectral IR Data in Temperature and Radiance Space