Ozone Profile and Tropospheric Ozone - PowerPoint Presentation

1. Ozone Profile and Tropospheric Ozone Retrievals from Joint UV and Visible Measurements by the TEMPOJuseon Bak, Xiong Liu, Chris Miller, Kelly Chance, John Houck, Ewan O’Sullivan, Caroline R. Nowlan, G. Gonzalez Abad @ Harvard-Smithsonian Center Benefits of visible fittinghelp distinguish boundary layer O3 from free tropospheric O3. Benefit of adding visible channel : better for detecting ground-level O3 (UV (290-340 nm) & VIS(540-650 nm) @0.6 nm, 0.2nm/pixel) Challenges of visible fitting 1. weak O3 absorption, strong interferences from surface reflectance and aerosols/clouds, other gases (O4, O2, H2O) 2. expensive RTM calculations 3. Need accurate radiometric calibration across the spectral range INTRODUCTION

2. O2: HITRAN 2016H2O: HITRAN 2016 O4:ThalmanO3: BDMOzone Profile and Tropospheric Ozone Retrievals from Joint UV and Visible Measurements by the TEMPO2.1 Strategy to account for gas interference and surface reflectanceParameterization w.r.t T    Cross Section Surface reflectance Gas Profile InputUpdated reflectance model improves -dependent reflectance predictions from MODIS MODIS Reflectance Fit Comparison over Los AngelesValidation of MODIS obs “simulated” from AVIRIS-NG obsNew probabilistic treatment removes spurious features Tropopause-based (TB) ozone profile climatology Hourly daytime ozone profile dataset from nature Run-GEOS-5 chemistry at 12 x 12 km2 (2012-13).Zonal mean climaTB climaOMI retrievals212 hPaInterpolation from pre-calculated LUTs w.r.t 17 Temp and 64 PressurebottomTopTropTBGEOS-ChemEOF (LAND)Polynomial (Snow/Water)Snow/asterWater/usgs- Jacobian w.r.t coefficient for each PC = =   =   =  [Zoogman et al., 2016, Chris Miller (in prep)]- Jacobian w.r.t coefficient for each poly = = As  MineralsArtifical MaterialsCoatingsSpilsVegetation     

3. Ozone Profile and Tropospheric Ozone Retrievals from Joint UV and Visible Measurements by the TEMPO2.2 Strategy for accurate, fast radiative transfer calculationTEMPO forward modelTempo spectrum: UV (290-340 nm) & VIS(540-650 nm) @0.6 nm, 0.2nm/pixel  = 800At what intervals should RTM should be run ? optimized intervals are found to have simulation errors less than 0.05 % = 13342 STEP1STEP2 Run fast PCA-RTM (4st, 24 layers, scalar)C  PCA-RTMLUT correctionaccurate/fast SSApproximate/fast MSCorrection factorVLIDORT run to get accurate/slow MS results, but For a reduced set of PCA-derived optical depths   Correcting to accuracy (12st, 72layers, vector)STEP3under-sampling corr 0.4 nm 305 nm0.1 nm 305 <  0.03 nmInterpolation/correctionABCApply to OMI o3p retRuntimeTCO-compare the existing (red/VLD) and updated (blue/PCA) Forward models; speed-up of 3.3 (runtime) and accurate-up of 5 % (TCO)Optimization of LUT correction to VIS channelScalar-vectorLow-high (32) streamCoarse-fine (99)layerVector, 48 layers, 12 streams are determined as accurate configuration ( Optimization of US correction to VIS channel C.1 RTM approximation errors (%) @ FWHM = 0.6 nm   C.2 VIS Correction Spectrum [w/o correction][with correction]DOptimized intervals (N=11933) vs 0.001 nm intervalsH2OO2586-600 nm@ 0.1 nm intervals N=140 @ 5nm @ 5nm626-632 nm@ 0.002 nm intervals N=3000

4. Cross SectionOzone Profile and Tropospheric Ozone Retrievals from Joint UV and Visible Measurements by the TEMPO2.3 Wavelength/Slit function/Radiometric CalibrationsSolar reference spectrum : S calibration  Super Gaussian-based slit func Soft Calibration-KNMI (Dobbe et al., 2008)-SAO (Chance and Kurucz, 2010)-CU-B(Odele Coddington, in prep)A.1 Example to OMI irradiance fittingA.2 Example to OMI radiance fitting-CU2020 looks the best candidate   =    =  Parameterization: slit paras are derived from irrad measurements through cross-correlation using solar reference with. 2. Linearization: impacts of slit function error on Spectral fit residuals are accounted for by including and  B.1 Example to OMIkind of systematic differences of characterized as a function of cross-track and wavelengths, or time using tropical cloud-free measurements  C.1 Example to OMPS for removing cross-track errorTCO w/o softTCO with softC.2 Example to OMI for removing degradation error no soft time-independent soft time-dependent softRMS of fit. residualsOMI – SONDE TCOApplying slit function linearization corrects errors due to slit function errors induced by different Solar reference. KNMICU-BSAO

5. Cross SectionOzone Profile and Tropospheric Ozone Retrievals from Joint UV and Visible Measurements by the TEMPO3. Summary and ConclusionImplemented the latest HITRAN 2016 databaseInvestigated the impact of applying different high-resolution solar spectrums on our retrievals Implemented Super Gaussian based slit function parameterization/linearization Merged Hourly-resolved ozone profiles taken from GEOS-chem model for the lower troposphere with Tropopause-based ozone profile climatology below as a priori dataImplemented PCA-RTM + LUT correction for a fast/accuracy RT simulation (UV only)Still need to include/optimize the following features - Evaluation of EOF based surface reflectance - Further speed up the RTM via LUTs for visible.Still need to demonstrate UV/Visible algorithm with real and GOME-2 and GEO-TASO/GCAS data