2004 ENVISAT Symposium Tropospheric Ozone and Ozone Profiles

2004 ENVISAT Symposium Tropospheric Ozone and Ozone Profiles from GOME Kelly Chance Harvard-Smithsonian Center for Astrophysics kchance@cfa. harvard. edu Xiong Liu, Christopher Sioris, Robert Spurr, Thomas Kurosu, Randall Martin, Michael Newchurch, P. K. Bhartia September 9, 2004

Outline • Introduction • Methodology • Validation with Dobson, TOMS, and ozonesondes • Global Distribution of tropospheric ozone • Summary and conclusions

Satellite-based Tropospheric Ozone Retrieval • Satellite observations are crucial for studying the global distributions, spatial and temporal variability, sources and sinks, transport, and seasonal behavior of tropospheric ozone. • Challenge: only about 10% of the total ozone, difficult to accurately separate tropospheric ozone and stratospheric ozone • Methods – Residual-based approaches: Total ozone – Stratospheric Ozone • • • – Coarse temporal resolution (i. e. , monthly) Subject to large uncertainties in the assumption made about stratospheric ozone Limited area coverage (e. g. , most of tropospheric ozone retrievals from TOMS are limited in the tropics) Direct ozone profile retrieval (e. g. , from GOME, OMI, SCIAMACHY, TES): forward model simulation + a priori knowledge + spectral fitting

Methodology n Optimal Estimation MIN n Measurements Misfit Smoothing and Regularization Ø GOME Channel 1 a and 2 b: 289 -307 nm, 327 -336 nm Ø n Spatial resolution: 960 km x 80 km Wavelength and radiometric calibrations Ø Derive variable slit widths and shifts between radiances/irradiances Fit shifts between trace gas absorption cross-sections and radiances On-line correction of Ring filling in of the solar and telluric absorption features l Channel 1 a: single scaling factor l Channel 2 b: 2 nd order poly. scaling factor to account for multiple scattering Ø Perform improved polarization correction using GOMECAL Ø Perform undersampling correction with a high-resolution solar reference Ø Perform on-line degradation correction for channel 1 a Ø Ø

Methodology n Improve forward model simulation LIDORT + look-up table correction of errors due to neglecting polarization Cloud-top height and cloud fraction from GOMECAT Monthly-mean SAGE stratospheric aerosols + GEOS-CHEM tropospheric aerosols Ø Daily ECMWF temperature profiles and NCAR/NCEP surface pressure Ø Initial surface albedo derived from 370 nm, which has minimal absorption Ø Wavelength dependent albedo (2 -order polynomial) in channel 2 b Ø Ø Ø n A priori Ø Ø n TOMS V 8 climatology [Mc. Peters et al. , 2003, AGU] Assume a correlation length of 5 km to construct a priori covariance matrix Retrieval Grid Ø Almost the same as 11 -layer Umkehr grid except the bottom 2 or 3 layers are modified by the NCAR/NCEP reanalysis tropopause pressure n State Vector: 50 parameters Ø 11 O 3 + 4 albedo (1 for ch 1 a & 3 for ch 2 b) + 4 Ring (1 for ch 1 a & 3 for ch 2 b) + 8 O 3 shift + 8 rad. /irrad. shift + 3 degradation correction (ch 1 a only) + 2 undersampling + 4 NO 2 + 2 Br. O + 2 SO 2 + 2 internal scattering

Validation: Hohenpeißenberg (48 N, 11 E) 19962000 Comparison with TOMS V 8 & Dobson total ozone, ozonesonde observation n Dobson/Ozonesonde: within ~8 hours, 1. 5° in latitude and ~600 km in longitude Ø Average TOMS total O 3 in GOME pixel Ø

Validation: Hohenpeißenberg (48 N, 11 E), 19962000 Ch 1 a degradation is well-handled GOME retrievals agree well with Dobson/TOMS and ozonesonde n n v v GOME-TOMS: 3. 2± 6. 0 DU GOME-Dobson: 4. 6 ± 11. 5 DU GOME-Ozonesonde Strat. : 2. 6 ± 13. 6 DU GOME-Ozonesonde Trop. : 0. 0 ± 6. 2 DU Average biases are within the range of ozone variability, retrieval and measurement uncertainties. n

Validation: Java (7. 6 S, 112. 7 E), 1996 -2000 Enhanced O 3 from biomass burning due to 97 -98 El Niño

Validation: Ascension (8. 0 S, 14. 4 W), 1997 -1999 Enhanced O 3 during biomass burning seasons

Validation: Ascension (8. 0 S, 14. 4 W), 1997 -1999 Total O 3 and tropospheric O 3 agree well with TOMS and ozonesonde n v v v GOME-TOMS: 1. 4± 3. 2 DU GOME-Ozonesonde Trop. : 1. 3 ± 8. 1 DU GOME-Ozonesonde Strat. : 14. 5 ± 9. 1 DU Large bias in stratospheric O 3. Thompson et al. [2002] also reported a ~8% TOMS-SHADOZ difference. n

Summary of comparisons for both ch 1 a and ch 1 a+ch 2 b retrievals

An Orbit of Retrieved Profiles (ch 1 a)

Global Distribution of Tropospheric Ozone High O 3 band North Africa Biomass Burning Low O 3 in the tropical Pacific Low O 3

Tropospheric Ozone (02/01/97 -02/28/97)

Global Distribution of Tropospheric Ozone High O 3 over NA and transport High O 3 from biomass burning and transport Low O 3 in the tropical Pacific Low O 3

Tropospheric Ozone (09/01/97 -09/30/97)

Summary and Conclusions • Ozone profiles and tropospheric ozone columns are derived from GOME using the optimal estimation approach after detailed treatments of wavelength and radiometric calibration and improvement of forward model inputs. • Retrieved total ozone compares well with TOMS and DOBSON total ozone. • The profiles, stratospheric ozone, and tropospheric ozone compare well with ozonesonde observations except for some stratospheric bias at the tropical stations. • Global distributions of tropospheric ozone are presented. They clearly show signals due to air pollution, biomass burning, and convection.

The End This work was supported by the Smithsonian Institution and NASA. We are pleased to acknowledge the cooperation of ESA and the DLR in making these studies possible.

Backup Slides

Validation: Lauder (45 S, 170 E), 1996 -2000

Validation: Lauder (45 S, 170 E), 1996 -2000 n GOME retrievals agree well with Dobson/TOMS and ozonesonde v v GOME-TOMS: -3. 1± 6. 8 DU GOME-Dobson: 0. 5± 17. 2 DU GOME-Ozonesonde Strat. : 3. 0 ± 9. 9 DU GOME-Ozonesonde Trop. : -0. 6 ± 6. 0 DU Average biases are within the range of ozone variability, retrieval and measurement uncertainties. n

Validation: Hilo (19. 5 N, 155 W), 1996 -2000 Dobson O 3 is measured at Mauna Loa (elevation: 3. 4 km)

Validation: Hilo (19. 5 N, 155 W), 1996 -2000 Total O 3 and tropospheric O 3 agree well with Dobson/TOMS and ozonesonde n v v GOME-TOMS: -1. 3± 4. 9 DU GOME-Dobson: 9. 3± 5. 6 DU GOME-Ozonesonde Trop. : -0. 3 ± 6. 6 DU GOME-Ozonesonde Strat. : 5. 9 ± 5. 9 DU Large bias in stratospheric O 3 especially after 1998. n

Validation: Samoa (14. 2 S, 175. 6 W), 1996 -2000

Validation: Samoa (14. 2 S, 175. 6 W), 1996 -2000 Total O 3 and tropospheric O 3 agree well with Dobson/TOMS and ozonesonde n v v GOME-TOMS: -2. 8± 3. 3 DU GOME-Dobson: 0. 6± 5. 6 DU GOME-Ozonesonde Trop. : 0. 8 ± 5. 7 DU GOME-Ozonesonde Strat. : 7. 6 ± 10. 8 DU Large bias in stratospheric O 3 especially after 1998, similar to another CMDL site Hilo. n

Validation: Java (7. 6 S, 112. 7 E), 1996 -2000 Total O 3 and tropospheric O 3 agree well with TOMS and ozonesonde n v v v GOME-TOMS: -4. 6± 3. 8 DU GOME-Ozonesonde Trop. : -0. 1 ± 5. 8 DU GOME-Ozonesonde Strat. : 10. 8 ± 8. 4 DU Large bias in stratospheric O 3. Thompson et al. [2002] also reported an ~8% TOMS-SHADOZ difference. n

Validation: Nairobi (1. 3 S, 36. 8 E), 1998 -2000

Validation: Nairobi (1. 3 S, 36. 8 E), 1998 -2000 n Total O 3 and tropospheric O 3 agree well with ozonesonde v v Dobson/TOMS and GOME-TOMS: -1. 1± 5. 1 DU GOME-Dobson: -1. 6± 5. 7 DU GOME-Ozonesonde Trop. : -2. 3 ± 7. 6 DU GOME-Ozonesonde Strat. : 8. 6 ± 6. 5 DU Large bias in stratospheric O 3, slightly larger than the ~2% TOMSSHADOZ difference. n