1 / 17

Institut für Umweltphysik (IUP) Institut für Fernerkundung (IFE)

CarbonSat L1L2 Study Final Presentation , 3-Jul-2013 . Universität Bremen, FB1 Physik und Elektrotechnik. Institut für Umweltphysik (IUP) Institut für Fernerkundung (IFE). Mission Requirement Consolidation: Spectral requirements, Gain-Matrix, Pseudo-Noise (PN)

sylvia
Télécharger la présentation

Institut für Umweltphysik (IUP) Institut für Fernerkundung (IFE)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CarbonSat L1L2 Study Final Presentation, 3-Jul-2013 Universität Bremen, FB1 Physik und Elektrotechnik Institut für Umweltphysik (IUP) Institut für Fernerkundung (IFE) • Mission Requirement Consolidation: • Spectral requirements, Gain-Matrix, Pseudo-Noise (PN) • ESA Study CarbonSat Earth Explorer 8 Candidate Mission • “Level-2 and Level-1B Requirements Consolidation Study“ • ESA Contract No4000105676/12/NL/AF M. Buchwitz, H. Bovensmann Institute of Environmental Physics (IUP) / Institute of Remote Sensing (IFE), University of Bremen (UB), Bremen, Germany 1

  2. Overview • This presentationcoversthefollowingtopicsrelatedtoLevel 1 Mission RequirementsConsolidation: • Gainmatrixapproachforrequirementformulation & verification • Illustratedfor „Relative SpectralRadiometricAccuracy“ (RSRA) requirement • Pseudo-Noisedue toinhomogeneousspectrometerentranceslitillumination • Spectralrequirements • Note thatseveralitemsrelatedtorequirementsconsolidationhavealreadybeenpresentedunder Agenda Item 3. • Other aspectsseefollowingpresentations: • RadiometricRequirements (UoL) • Spatial / temporal co-registration (SRON) • High spatialresolutionsamplingfor C&A correction/flagging (IUP) 2

  3. Gain Matrix (GM) approach Gain Matrix (GM) approach forrequirement formulationandverification 3

  4. Gain Matrix (GM) approach: The issue • The initial (e.g., MRD v1.1) Relative SpectralRadiometricAccuracy (RSRA) requirement • formulation was „simple“ („< 0.05% (T) withineach band“) but • verydifficulttobemetand • includedspectralregions, where relaxed valueswould also beappropriate („overspecification“) • Itthereforehasbeeninvestigatedifthe RSRA requirementcanbereformulatedusing a Gain Matrix (GM) formulation 4

  5. Gain Matrix (GM) approach • GM approach (detailssee TN-4b): • ComputeGain Matrix G (how?: via L1->L2 retrievalalgorithm) • ComputeΔx = G Δy, where • G = Sx KT Sy-1 , withSx = (KT Sy-1 K + Sxa-1)-1 • Δyisthemeasurementerror#) (a spectrum = a vector) and • Δxisthesystematicerror (vector) ofretrievedgeophysicalparameters, whichcanbeconverted (using simple formulas) totheerrors (biases) oftheparametersofinterest, i.e., ΔXCO2andΔXCH4 • Finally, onehastocomparetheresultingΔXCO2and ΔXCH4valueswiththepermittedmaximumerrorfortheerrorsourceforwhichΔyhasbeencomputedtodetermineifthecorrespondingrequirement (e.g., RSRA) ismetornot. • #) relative errorofthemeasuredreflectance, or, moreprecisely: in thecorresponding TN, Δyisdefinedastheratioof an „erroneous“ reflectanceandthecorrespondingerror-freereflectance; however, adding +1.0 hasessentiallynoeffectbecauseofthepolynomialused in theretrievalalgorithm 5

  6. GM: Status TN-4b • TN-4b describes GM approach & GMs asdeliveredto ESA • GM approachusedfor MRD v1.2 • But: Refinements still ongoing: Latestdeliveryof IUP->ESA: Gs + correspondingreflectancesfor so-calledLdarkandLbrightscenes 6

  7. Pseudo Noise Pseudo-Noise (PN) due toinhomogeneousspectrometerentranceslitillumination 7

  8. Pseudo noise (PN): Approach • A PN datasethasbeenprovidedby ESA: Contains PN spectrafor 7 scenarios (IH0..IH6) • Foreach band and all 7 scenariosreflectanceratio „errorspectra“ (Δy) havebeengeneratedcontaining INHOM / HOM reflectanceratios: Δy SWIR-1 NIR SWIR-2 All y-scales: +/- 2% Reflectanceratios: upto +/-2% for NIR & SWIR-2; muchsmallerfor SWIR-1 8

  9. PN: Error estimates:Worstcasefor XCO2 Δy 9

  10. PN: Error estimates:Worstcasefor XCH4 Δy 10

  11. Pseudo noise (PN): Summary andconclusions • Resultsarebased on simulatedretrievalsappliedtoreflectancespectraperturbedby „typical“ PN errors: • Findings: • Withoutanycorrectiontheerrorscanbequitelarge • Using SHIFT & SQUEEZE theerrorsaresignificantlyreduced • These errorvaryfromgroundpx-to-pxandaretherefore „noise“ ratherthanspatio-temporallycoherentbiases on the relevant spatio-temporal scales-> Impact on precision (= randomerror) • Analyzing~1400 scenariosthefollowinghasbeenfound: • The largestsinglescenarioerrorsfoundare1.3 ppm for XCO2and9 ppb for XCH4. Forthisworstcase: • XCO2precisiondegradation: 1 ppm (T) -> 1.6 ppm (=√(12+1.32)) • XCH4precisiondegradation: 10 ppb (T) -> 13.5 ppb ((=√(102+92)) 11

  12. SpectralRequirements SpectralRequirements 12

  13. Spectralstability / knowledge Spectralstability / knowledge MRDv1.1 MRDv1.1 • T: 0.05 SSI = SSI/20 = FWHM/60*): • NIR: FWHM = 0.1 nm -> 0.002 nm • SW1: FWHM = 0.3 nm -> 0.005 nm • SW2: FWHM = 0.55 nm -> 0.009 nm • *) assuming: SSR=3 Welljustifiedorisitpossibleto relax thisrequirement? 13

  14. Spectralerrors: Case 3: Squeezeerrorwithin FWHM/10 Biases • Ifusing SH&SQ anditeration: • XCO2: OK • XCH4: OK 14

  15. Spectralerrors: Case 6: Quadraticerrorwithin FWHM/20 Biases • Ifusing SH&SQ anditeration: • XCO2: large errors • XCH4: large errors 15

  16. Spectralerrors: Summary & conclusions • Withoutcorrection XCO2and XCH4biasescanexceed (violate) theaccuracyrequirementifspectralerrorsreachorexceedthe T req. (< 0.05 SSI = FWHM/60) • With „shift & squeeze“ correctionerrorsaresignificantlyreduced • However, errorsareonly „small“ if „shift & squeeze“ is a goodmodelforthespectralerrors (in thiscase larger spectralerrorsareacceptable, e.g., FWHM/30, possiblyeven FWHM/20; thisrequireshoweverseveraliterations); ifthisis not true, e.g., becauseof „quadraticerrors“ ormorecomplexspectraldependencies, thebiasesmaybe larger thanrequired. • Iftheerroriswelldescribedby „shiftandsqueeze“, therequirementcanbe relaxed by (at least) a factorof 2. Ifthisis not true, therequirementcannotbe relaxed. • Not discussed but „obvious“: In-flightstabilityisveryimportant(evenmorethanknowledge) becausewavelengthcalibrationcanbeimprovedwith L1-2 tools (bycarefullyinvestigating sub-sets oftheobs.) andiftheerrorsarestable „all“ datacanbereliablyprocessedusingtheimprovedcalibration – otherwiseveryfrequentspectralcalibrationisneeded. 16

  17. Summary andConclusions • Specificconclusions: • See „Summary andConclusions“ givenforeachtopic • General conclusion: • Initial requirements (e.g., MRD v1.1) canbejustified – at least for „worstcase“ situations • Not clearhoweverhowrealistictheassumederrorsare !? • Itisthereforeimportanttorepeatthisanalysiswithrealistic instrumental / residual calibrationerrors • The resultshavebeenusedtoreformulatethecorrespondingrequirements (-> MRD v1.2) 17

More Related