1. Intercomparison of SCIAMACHY NO2, the Chimère air-quality model and
surface observations
Nadège Blond,
LISA, Paris, France
Henk Eskes, Folkert Boersma,
Ronald van der A
KNMI, Netherlands
Michel van Roozendael,
Isabelle De Smedt
BIRA-IASB, Belgium

2. Slant column retrieval approach (BIRA-IASB)
DOAS slant column:
• "raw" L1, v 4.02 L1, v 5.01 L1
• Non-linear least-squares inversion (Marquardt-Levenberg)
• Wavelength window
• NO2 243K (Bogumil), O3 (Bogumil), O2-O2, H2O
• 2nd order polynomial
• Undersampling cross section
• Ring (Vountas)
• Offset correction based on measurement over Indian Ocean

3. Combined retrieval - modelling - assimilation approach to SCIA NO2
Careful treatment needed for:
• Clouds
• Surface albedo
• Profile shape
• Aerosol

4. Slant to vertical column retrieval approach (KNMI)
Air-mass factor calculation:
• Temperature correction (NO2 cross section)
• TM3 / TM4 (tropospheric) CTM
• Assimilation of slant columns -> stratospheric "background"
• Fresco cloud fraction and cloud top pressure
• TOMS / GOME combined albedo map (Herman, Koelemeijer)
• DAK RTM height-dependent AMF lookup table
• Tropospheric AMF based on TM profile shape, clouds
Product:
• Detailed error estimates
• Averaging kernels

5. Validation results (ACVE-2), stratosphere
J. C. Lambert
NO2 products:
• SCIA processor
• IUP
• SAO
• BIRA-IASB
• Heidelberg

6. Combined retrieval - modelling - assimilation approach
to GOME NO2

7. Chimère model
Developed in France
R. Vautard, H. Schmidt, L. Menut, M. Beekman, N. Blond, ... )
Operational air-quality forecasts:
Model ingredients:
• MELCHIOR chemistry (82 species, 333 reactions)
• EMEP emissions
• ECMWF meteorological analyses
• 15 vertical layers, surface hPa
• Boundary conditions from MOZART monthly-mean climatology

8. Emissions

9. Intercomparisons Chimère, SCIA and surface observations
Motivation:
• Lack of profile observations of NO2 for validation purposes:
use model as intermediate for indirect validation study
Approach:
• Space-time collocation of Chimère fields to individual SCIA pixels
• Application of averaging kernels:
Simulated SCIA-equiv column = kernel vector • model NO2 profile
• One year of SCIA data, 2003; Cloud free (cloud radiance < 50%)
Advantages:
• Compare model-SCIA under exactly same conditions (e.g. cloud free)
• Comparison independent of profile shape assumptions in the
retrieval

10. Chimère and surface observations (RIVM, NL)
Netherlands:
(rural stations)
Bias 0.1 ppb
RMS 7.2 ppb
Correl. 0.66

11. SCIAMACHY vs. Chimère: yearly mean
Yearly-mean bias = molec cm-2, RMS 2.9, correl.coeff. 0.73
Cloud-free pixels

12. SCIAMACHY vs. Chimère: 27 Feb 2004

13. SCIAMACHY vs. Chimère: 28 Mar 2004

14. SCIAMACHY vs. Chimère: 16 April 2004

15. SCIAMACHY vs. Chimère: 16 Sep 2004

16. Synergy: Surface - Chimère - SCIAMACHY

17. Conclusions NO2 comparisons
SCIAMACHY - Chimère - surface
• Yearly mean:
- very small bias SCIA - Chimère and
Chimère - surface
- Correlation coefficients 0.7 typically
• SCIA and Chimère resolution comparable
• Extended NO2 plumes compare well
• Details show differences:
- Seasonality (winter Chimère higher)
- Individual days
- Distribution
- Amount of detail