The CONCORDIASI Workshop, Toulouse, March 2010 Impact study of AMSU-A/B data over land and sea-ice in the Météo-France global assimilation system Fatima Karbou and Florence Rabier CNRM-GAME, Météo-France & CNRS

The CONCORDIASI Workshop, Toulouse, March 2010 Assimilation of AMSU-A & AMSU-B over land Assimilation of AMSU-A & AMSU-B over land Since July 2008, a dynamical retrieval method is used in ARPEGE to estimate the land surface emissivity at microwave frequencies (Karbou et al. 2006): Instantaneous emissivity retrieval at one surface surface channel (89 GHz for AMSU-B and 50 GHz for AMSU-A) The emissivity is then given to sounding channels (with no frequency dependency) Top of Atmosphere Energy source Surface (emissivity, temperature) (1) Upwelling radiation (2) Downwelling radiation (3) Surface emission Signal attenuated by the atmosphere Plane parallel non scattering atmosphere, specular surface

The CONCORDIASI Workshop, Toulouse, March 2010 Indirect measurements of temperature and humidity AMSU-A & AMSU-B observations AMSU-A & AMSU-B observations

The CONCORDIASI Workshop, Toulouse, March 2010 More humidity in EXP TCWV (EXP-REF) Correlations with GPS Evaluation wrt GPS data from AMMA Assimilation of AMSU-B (low peaking humidity channels) over land

The CONCORDIASI Workshop, Toulouse, March 2010 More humidity in EXP TCWV (EXP-REF) Diurnal cycle of TCWV, Timbuktu (MALI) Assimilation of AMSU-B over land operational in April 2010

The CONCORDIASI Workshop, Toulouse, March 2010 Current usage of AMSU-B channel 5 ( GHz) in ARPEGE, dec 2008 One of the limitations: large uncertainties about the surface description (emissivity and surface temperature) over snow and sea-ice AMSU-A & AMSU-B observations AMSU-A & AMSU-B observations over sea-ice

The CONCORDIASI Workshop, Toulouse, March 2010 Assimilation of observations over sea-ice July 2009January 2009 Difficult modelling of sea-ice emissivity OPER Emis=0.99 Emissivity at 89 GHz

The CONCORDIASI Workshop, Toulouse, March 2010 (a) For AMSU-B in particular, can we still use the 89 GHz emissivities for sounding channels without any frequency dependence parameterization ? (b) Can we safely use the specular assumption over sea ice ? January 2009July 2009 Assimilation of AMSU-A & AMSU-B over sea-ice Assimilation of AMSU-A & AMSU-B over sea-ice

The CONCORDIASI Workshop, Toulouse, March 2010 (a) For AMSU-B in particular, can we still use the 89 GHz emissivities for sounding channels without any frequency dependence parameterization ? (b) Can we safely use the specular assumption over sea ice ? Introduction of frequency parameterization for sea ice: to describe the emissivity change from 89 GHz to GHz (AMSU-B) Emissivity (~183 GHz) = Emissivity at 89 GHz + f (Tb 89, Tb150, Ts) (AMSU-A) Emissivity (~52-60 GHz) = Emissivity at 50 GHz Data impact studies for evaluation: Period: 15/12/2009 to 04/02/2010 CTL: the operational system EXP: CTL + emissivity model over sea ice + assimilation of AMSU-A/-B over sea ice Assimilation of AMSU-A & AMSU-B over sea-ice Assimilation of AMSU-A & AMSU-B over sea-ice

The CONCORDIASI Workshop, Toulouse, March 2010 Data impact results Data impact results Usage of AMSU-B channel 5 ( GHz) in ARPEGE CTLEXP

The CONCORDIASI Workshop, Toulouse, March 2010 Usage of AMSU-B channel 5 ( GHz) in ARPEGE CTLEXP Snow (effect of specular approximation ?, Harlow, 2009, Guedj et al. 2010) Data impact results Data impact results

The CONCORDIASI Workshop, Toulouse, March 2010 Assimilation of observations over sea-iceOperational in Nov 2010 Wind scores wrt radiosondes (January 2009)

The CONCORDIASI Workshop, Toulouse, March 2010 Assimilation of observations over sea-ice Scores: Plutôt positifs en particulier sur lHémisphère Nord Scores: Plutôt positifs en particulier sur lHémisphère Nord

The CONCORDIASI Workshop, Toulouse, March 2010 Summary Objective: extend the use of AMSU observations over sea ice Method to calculate the sea ice emissivity to be used to assimilate humidity and temperature observations Beneficial for ARPEGE: data usage, RTTOV performances, Fit to all available observations, quality of analyses/forecasts Issues: the use of AMSU data over snow surfaces