1. Data assimilation of atmospheric CO 2 at ECMWF in the context of the GEMS project Richard Engelen ECMWF Thanks to Soumia Serrar and Frédéric Chevallier

2. Global and regional Earth-System (atmosphere) Monitoring using Satellite and in-situ data (GEMS) T. Hollingsworth 1, O. Boucher 2, H. Eskes 3, C. Granier 4, V.H. Peuch 5, P. Rayner 6, M. Schultz 7, A. Simmons 1 1 ECMWF; 2 UK Met Office; 3 KNMI, NL; 4 Service d'Aéronomie CNRS, 5 Meteo-France; 6 Laboratoire des Sciences du Climat et de l‘Environnement; 7 Forschungszentrum Jülich, Germany

3. Prime Objectives Creation of an pre-operational global monitoring system for greenhouse gases, reactive gases, and aerosols in the troposphere and in the stratosphere by early 2009. This system will be capable of producing near-real- time (depending on data availability) and retrospective global analyses for monitoring atmospheric composition, and short-range forecasts to drive regional air-quality models. It will provide information relevant to the Kyoto and Montreal protocols, and to the UN Convention on Long-Range Trans-boundary Air Pollution.

4. Greenhouse Gases (GHG) Global Reactive Gases (GRG) Aerosols (AER) Regional Air Quality (RAQ) GEMS themes Production (PRO) & Validation (VAL) GHG GRG AER RAQ

5. Greenhouse Gases (GHG) GHG Theme coordinator: Peter Rayner, LSCE, France Development of an operational system to use satellite and in-situ data to monitor the concentrations of greenhouse gases, and their associated surface sources and sinks, and to attribute these sources and sinks to controlling processes. Carbon fluxes from the terrestrial biosphere on August 10, 2003 (heat wave in Europe and drought in the USA) as simulated by the model ORCHIDEE [g-C/m2/day]

6. 2-step process 4D-Var atmospheric data assimilation (12 hour assimilation window) AIRS IASI OCO GOSAT 4D-Var flux inversion (1 year inversion window) Updated fluxes Flasks Tall towers Prior fluxes

7. Terminology Free running CO 2 model: CO 2 is transported using constrained meteorology from operations every 12 hours. Reanalysis/assimilation: CO 2 is constrained by AIRS, while meteorology is constrained by all other available operational observations at the same time. Fluxes used in the model CASA diurnal natural biosphere Takahashi ocean Andres anthropogenic GFED2 fire emissions

8. CO 2 reanalysis with AIRS Monthly mean CO 2 column-mean mixing ratios, after 8 months of assimilating AIRS radiances, show small but significant changes to a simulation with free-running CO 2. After validation these differences will be used to estimate modifications in the prior surface flux fields. Analysed CO 2

9. The AIRS radiances are also able to adjust the shape and amplitude of a CO 2 plume resulting from extensive biomass burning over Africa. The GEMS monitoring system will have the capability to follow these plumes in near- real time in terms of CO 2, CO, and aerosol concentrations, constrained by various satellite observations.

10. Early Validation Comparisons with flight data over Hawaii (courtesy of Pieter Tans, NOAA/ESRL, shows a clear improvent of the analysis over the free- running model. The top plot shows a best- case individual comparison (11 May 2003). The bottom plot shows an average over all available flight profiles between 1 January and 30 October 2003.

11. Molokai, Island, Hawaii Blue: free-running model Red: reanalysis Black: observations

12. Molokai, Island, Hawaii Blue: free-running model Red: reanalysis

13. Surgut, Siberia courtesy of Dr Machida, NIES

14. Surgut, Siberia

15. Molokai, Island, Hawaii

16. Individual Profiles 10 ppmv 11 May 2003 9 October 2003

17. There is a summer trend in atmospheric CO 2 that is not corrected by the AIRS observations at the moment. Possible solutions: Increase background errors to allow bigger adjustments from the AIRS observations Apply a model bias correction to keep the problem as un-biased as possible. This has to be accounted for in subsequent flux inversions. Use optimized fluxes instead of CASA Issues: Missing sink is not modelled and therefore cannot take out excess CO 2 from the atmosphere The background constraint is too strong The AIRS bias correction might not work properly AIRS might not provide a strong enough constraint to adjust the concentrations

18. Conclusions GEMS is well on its way to build an atmospheric composition monitoring system First 1-year CO 2 reanalysis has been run using AIRS observations Potential problems are being identified and where possible corrected before a new run will be started Results will be provided to inverse flux model at LSCE

19. Molokai, Island, Hawaii Blue: without fire emissions Red: with fire emissions