1, J.-L. Attié 2, J.-P. Cammas 2, A. Dufour 1, B. Josse 1, F. Karcher 1, S. Massart 3, M. Michou 1, T. Morel 3, P. Nédélec 2, S. Pradier 2, P. Simon 1, H. Teyssèdre 1 and D. Cariolle 3 1.Météo-France, Centre National de Recherches Météorologiques 2.Laboratoire d’Aérologie, CNRS-UPS, Toulouse 3.CERFACS Acknowledgments to the MOZAIC and MOPITT teams Chemical assimilation and forecasts with the multiscale CTM MOCAGE Chemical assimilation and forecasts with the multiscale CTM MOCAGE
MOCAGE growing number of real-time applications in connection with atmospheric environment : monitoring protocols (Montreal, Kyoto, CLRTAP), air quality and factors affecting health (UV, pollens,…), accidental releases, campaign planning. These applications occur at several scales and may concern the stratosphere, troposphere as well as PBL. They share many aspects (advection, parameterizations,…). MOCAGE (Model of atmospheric chemistry at large scale) has been developped since 1999 with the objective to complement Météo-France Numerical Weather Prediction operations in the mentioned fields. It is now a running prototype for GMES applications, inclunding real-time forecasting and chemical data assimilation.
~ 10 6 gridpoints, 150 tracers (gaz, aerosol bins) ARPEGE 2° ARPEGE 0.5° ALADIN 0.1° L47 configuration for forecasts
ENSEMBLE ex. 6 (JRC) : MOCAGE (crosshatch) vs other models Release (fictive) : Dublin, 25/06/ h UTC, lasting 15h Plume tracking : 60 hours
210-Pb validation Petit-Saut, French Guyana (gray, LGGE-CNRS)
MOCAGE J Up to 96h / 3h J+1 ARPEGE-ALADIN NWPM MOCAGE Daily 96h multiscale forecasts
2003 air quality forecasts verification in France (ozone, J forecast)
Skill score on ozone (200 sites, bias in g/m 3 )
Ozone hole forecasts in 2002 and 2003 (no chemical data assimilation : free forecast )
UV index forecast (w. clouds) (28/08/02) A posteriori check of NWP cloudiness Phenomenological correct. in case of mismatch
Ozone evaluation in the UTLS : MOZAIC
35°N How to have an impact on global UTLS fields with only 5-10 flights per day? [Cathala, Pailleux and Peuch, Tellus, 2003] non-local assimilation uses a (ozone equivalent latitude, ) coordinate system, similar to (PV EL, ) [Lary et al.,1995], but avoids small scale numerical structures in PV. Idea : shapes of ozone isolines are OK, but assimilation could modify gradients a few flights per day allow to constrain the Northern hemisphere UTLS ozone, as assessed by independent ozone sounding.
MOZAIC ozone MOCAGE MOCAGE + assimilation MOZAIC flight level pressure One year (2000) with and without assimilation
Assimilation of MOZAIC data allows to drive long model integration with « constant » quality in the UTLS Feb. 2000April 2000June 2000 MOCAGE MOCAGE + assimilation SONDES MOCAGE
Impact of MOZAIC assimilation Mean vertical profile for the zonal band (20°N-60°N) MOCAGEMOCAGE + assimilation
MOPITT CO assimilation NCAR Kalman-Bucy filter Relative diff. (%) MOCAGE / MOPITT 500 hPa 11-30/03/2001 Relative diff. (%) MOC. + ASSIM / MOPITT 500 hPa 11-30/03/2001
Evaluation against independent data Top : CMDL & TRACE-P flights Left : one CMDL profile (11/03/ °N, 157°W )
MOCAGE & EU ASSET project
GOME ozone 3d-FGAT ( KNMI GOME ozone profiles retrievals ) Before assimilation After assimilation
Overview MOCAGE is one of the first atmospheric chemistry multiscale strat.-top. suite used for daily 96h forecasts. Data assimilation (in situ : surface, MOZAIC profiles and flight level measurements ; space : GOME, MOPITT, ENVISAT) is now in a research mode. Main problems are now « accurate » emissions specification and NRT data access. MOCAGE is involved in : MOZAIC-III, PRISM, ASSET, SCENIC (FM5) ; ACCENT, SCOUT (FM6) and the proposals GEMS and BACC-TO-BACC. Operations : national air quality (Prév’Air / platform with the CHIMERE model, CNRS) ; UV forecasts ; accidental releases