NNORSY-GOME Ozone Profile Retrieval Products and Climatology A.K. Kaifel, J.S. Kaptur (ZSW), J.C. Lambert, C. de Clercq (BIRA-IASB), T. Erbertseder, F. Baier (DLR-DFD) J.M. Krijger, I. Aben (SRON) Center for Solar Energy & Hydrogen Research (ZSW), Stuttgart, Germany Belgian Institute for Space Aeronomy (BIRA-IASB), Brussels, Belgium German Aerospace Center (DLR), Oberpfaffenhofen, Germany Space Research Organization Netherlands (SRON), Utrecht, Netherlands funded by ESA under Contract: 17892/03/I-LG ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Ozone Profile Retrieval Products and Climatology Outline 1. NNORSY and Optimal Estimation (OE) 2. Approach for NNORSY 3. Local error estimation 4. NNORSY-GOME ozone profile products 5. Validation: Geophysical and data assimilation 6. Application for NNORSY-GOME data set 7. Dynamic ozone profile climatology 8. Summary ITSC-16, May 2008, Angra dos Reis, Brazil Optimal Estimation z.B. FURM, OPERA, ... x ~1 min/profile x RTM y "a priori" y GOME spectrum x p(x|y) = p(y|x) p(x) p(y) O3 profile … … NNORSY Comparison of retrieval approaches Optimal Estimation ~1 ms/profile k(x,y) y x collocated measurements NNORSY ITSC-16, May 2008, Angra dos Reis, Brazil Overview NNORSY-GOME Operation GOME profiles operational real-time processing network param.s quality check GOME Level 1 data SAGE, POAM, HALOE profiles neural network training collocation collocated spectra/ profiles collocation SHADOZ sondes collocation WOUDC sondes Preprocessing training data test data eval. data random data mixing Training ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Input Parameters Input data for neural network yi GOME spectral data geolocation meteorol. info climatological predictors 270 – 325 nm (O3 Hartley & Huggins) 380 – 385 nm (atmosph. window) 598 – 603 nm (O3 Chappuis) 758 – 772 nm (O2→ cloud info) SZA, SAA sensor scan angle LOS type (east, nadir, west) GEOS4 temperature profile Latitude Season and sensor age NNORSY (Neural Network Ozone Retrieval System) ITSC-16, May 2008, Angra dos Reis, Brazil Estimation of Local Retrieval Error training data 1 Example for NNORSY Ozone Retrieval y t NNORSY retrieval training data 2 1 y o |o-t| t error network training NNORSY error estimation ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Processing Model NNORSY ozone profile retrieval output data GOME Lv. 1 ESA GOME Lv1 file list GOME Level 1 ZSW Gdp_ex GOME Lv1 file list 1.2 TB GEOS4 T-profile NNORSYGOME processing GDP4 TOZ TOMS TOZ Lv3 TPH GEOS4 Lv3 Optional auxiliary Input data NN error coeff. file NN index file Neural network input parameter files NN retrieval coeff. file Lv2 profile Level 2 ozone profile data TOZ Level 3 total ozone column TCO Level 3 tropospheric Ozone column SCO Level 3 stratospheric Ozone column TOR GDP-SCO Level 3 tropospheric ozone residual using GDP4 data TOR TOMS-SCO Level 3 tropospheric ozone residual using TOMS V8 data Trop. mean VMR Mean tropospheric ozone volume mixing ration ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Ozone Profile Data Set Temporal coverage Additional Data Start Date: 09/01/1995 Pressure: hPa End Date: 05/31/2003 Total Ozone: Integr. NNORSY-profiles & TOMS (DU) Number of Orbits/Profiles: 36694/48.579.769 Tropopause Height: km Level 2 Profiles Tropospheric Ozone: DU Minimum Altitude: surface Maximum Altitude: 61 km Vertical Sampling: 1 km Stratospheric Ozone: DU GEOS4 T-profile GEOS4 surface pressure Ozone Number Density: 1e18/m³ Ozone Error: % ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Single Ozone Profile I ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Single Ozone Profile II ITSC-16, May 2008, Angra dos Reis, Brazil GOME Instrument Degradation Ratio between model and measurement (yearly average) Mg Solar lines Mission start ITSC-16, May 2008, Angra dos Reis, Brazil Correlative Analysis: Artic ITSC-16, May 2008, Angra dos Reis, Brazil Correlative Analysis: 400 S / New Zealand ITSC-16, May 2008, Angra dos Reis, Brazil Validation with Data Assimilation Why ¾Global quality control by first guess and error statistics ¾Identification of outlier or systematic errors ¾Check of retrievals for areas of known GOME problems (SAA) ¾Grid parameter: CTM Model ¾Based on ROSE/DLR assimilation system, Version 3.0 (Baier et al., 2006) ¾Horizontal optimum interpolation scheme with analyzed error (Kalman Filter) ¾Comprises all relevant chemical and dynamical processes of stratosphere ¾DA system driven by UK Met Office wind and temperature fields ¾Grid parameter: ¾Vertical coverage: 1000 – 0.1 hPa ¾Vertical discretisation: 43 Levels ¾Horizontal resolution: 2.5° x 3.75° resolution ITSC-16, May 2008, Angra dos Reis, Brazil Data Assimilation Overview Input = NNORSY-GOME retrievals Forecast CTM First Guess NNORSY OI Profile Correction Output = 3D global ozone analysis Diagnostic parameters derived OMF = observation minus first-guess AMO = analysis minus observation AMF = analysis minus first-guess ITSC-16, May 2008, Angra dos Reis, Brazil Validation with Data Assimilation Zonal Means of First-Guess minus Observation Error (FMO) January, April, July, October ITSC-16, May 2008, Angra dos Reis, Brazil Validation with Data Assimilation Daily zonal means of FMO Error at 100 hPa ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Application: Ozone Reanalysis for Climate Model Evaluation (1/2) GOME/NNORSY reanalysis vs ECHAM E39/C 1996 1999 1996 1999 100 hPa 50 hPa 30 hPa 10 hPa 0 2 4 6 8 ppm(v) 0 2 4 6 8 ppm(v) ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME Application: Ozone Reanalysis for Climate Model Evaluation (2/2) NNORSY-GOME reanalysis vs ECHAM E39/C 0 2 4 6 8 ppm(v) 0 2 4 6 8 ppm(v) ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY Dynamic Ozone Profile Climatology CLIMATOLOGY-CALCULATION INPUTPARAMETER profile ozone stdv temperature stdv profile MODE 1 #1 #2 #3 #4 MODE 2 #5 #6 #3 #4 MODE 3 #7 #8 MODE 4 #9 #10 Time TOZ T-profile ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY Ozone Profile Climatology Example for single profiles: Tropics, Antartica ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY Climatology vs. Retrieval ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY Climatology vs. Ground Based (1/2) ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY Climatology vs. Ground Based (2/2) ITSC-16, May 2008, Angra dos Reis, Brazil NNORSY-GOME: Summary ► Almost 10 year global NNORSY-GOME ozone profile data set ► Geophysical validation and comparison with sonde/lidar/satellite ► Validation and global consistency check with data assimilation ► Various tropospheric ozone column products ► NNORSY compensate for instrument degradation ► New dynamic ozone profile climatology ► All products and data are available: http://nnorsy.zsw-bw.de ► Extending GOME ozone profile data record with SCIAMACHY ► NNORSY application to SCIAMACHY nadir ITSC-16, May 2008, Angra dos Reis, Brazil THANK YOU UKHYANTY NNORSY-GOME: The End QUESTIONS ITSC-16, May 2008, Angra dos Reis, Brazil