F. Rabier
C. Faccani, N. Fourrié, É. Gérard, V. Guidard,
F. Karbou, P. Moll, R. Montroty, C. Payan, P. Poli
Météo-France/CNRS, Toulouse, France

Stretched grid, 4D-Var assimilation (6-hour window)

AMV with QI
MODIS polar winds
SSM/I
AIRS
Ground-based GPS over Europe (E-GVAP)
Scatterometers
ATOVS from MetOp
MSG CSR
AIREP density
GPS radio-occultation
IASI
Variational Bias Correction for radiances
Emissivity parametrisation over land for micro-wave
Tropics, Poles

Data
Methods
GPS radio-occultation
IASI
Variational Bias Correction for radiances
Emissivity parametrisation over land for micro-wave

Assimilation of GPS - Radio Occultation data

Data assimilated
–
Bending angles
COSMIC/CHAMP/GRACE
– From 1 to 18 km altitude
– In ARPEGE 4DVAR
 1D- Observation operator from
ECMWF

Developments :
– Data selection algorithm to avoid radio propagation pbs, independent from background check
–
Thinning, observation errors…
Poli et al, 2008
Paul Poli
6-hour data coverage
DPrevi/COMPAS

GPS radio-occultation impact on the forecast:
Scores wrt RS, 41days, March-April 2007
Diff. RMS Geopototential Diff. RMS Temperature
Paul Poli
BLUE = GPSRO better
RED = GPSRO worse
Diff. RMS Wind

 Level 1C radiances are received via EumetCast in Toulouse
(whole BUFR including 8461 channels)
 A subset of 314 channels is monitored (commonly chosen with other NWP centres)
 Radiances are bias corrected
 Data selection:
– Cloud detection is based on a channel ranking method from ECMWF McNally & Watts (2003)
– 50 channels are actively assimilated, only over sea
(peaking between 100 hPa and 620 hPa)
Vincent Guidard, Nadia Fourrié

IASI is currently assimilated in e-suite, with positive impact (esp. SH). Similar impact was obtained in a summer trial
IASI +
IASI -
RMS error difference ,Geopotential, 72h range, 22 days
Vincent Guidard

 Satellite radiance data have systematic biases that can depend on the scan angle
(geometry) and on the flow
They can be explained by predictors such as powers of the scan angle, thicknesses of some layers of the atmosphere, skin temperature, etc., by a mulitple linear regression
 In the VarBC scheme, coefficients of the regression are dynamically adapted at each analysis time.
They are included in the control variable of the assimilation, and they use other
''conventional'' data (like radiosondes or aircraft data) as a constraint.
Dee (2004) ,
Auligné et al (2007)
 Channel #8 of AMSU-A onboard MetOp-A
 Learning period: 5 to 15 days
Vincent Guidard, Elisabeth Gérard

 Impact over a 43-day period of July-August 2007
VarBC for AMSU-A/B, MHS, SSM/I, HIRS & AIRS, versus static bias correction
Impact on RMSE wrt radiosondes for geopotential height
Very positive !
Vincent Guidard, Elisabeth Gérard

 AMSU microwave observations
– Great potential for estimating atmospheric temperature and humidity over all surfaces
 Over the ocean
– Emissivity ~ 0.5
– Emissivity models good enough to meet the NWP requirements
– The sea surface contribution to the signal is lower than the land contribution
 Over land
– High emissivity (~1.0)
– Only channels that are the least sensitive to the surface are currently assimilated
– Remaining large uncertainties on land emissivity and skin temperature
• difficulties to describe the emissivity variation in time and space and with surface types, roughness and moisture content
 Recent advances at Météo-France in the estimation of land emissivity and skin temperature from satellite microwave observations
– To help surface and sounding channel assimilation
– Statistic to dynamic approaches
– Experiments with AMSU and SSM/I
Karbou et al, 2006
Fatima Karbou, Elisabeth Gérard

Observed Tb at polarization p and frequency 
Hypotheses:
• plane parallel non scattering atmosphere
• specular surface
T ( p ,

)
 
( p ,

).
T s
.
 
( 1
 
( p ,

)).

.
T (

,

)

T (

,

)
Emissivity
Energy source
Outputs of RT model (RTTOV):
T & q short-range forecasts
Top of Atmosphere
Signal attenuated by the atmosphere
Surface (emissivity, temperature)

Number of assimilated data, AMSU-A channel 7, August 2006
Reference
Emissivity dynamically estimated from Ch3
14
Number of data used
Fatima Karbou
Number of data used

Towards the assimilation of surface channels?
Scores geopotential, 36 days, summer 2006, wrt radiosondes
Bootstrap, ++ : exp better than Ref (95% confidence level)
Assimilation experiment: using channels 2 & 5 AMSU-B over land
Fatima Karbou

Tropics: AMMA, Tropical cyclones
Antarctic: Concordiasi

 Field experiment in 2006
 Extra RS stations in red
 First data impact study without any special bias correction,
Others to be performed in red
Claudia Faccani, Patrick Moll

Improvement for high-level peaking channels (10 to 13)
Domain:28N, 8S, 30W, 35E
Claudia Faccani

AMMA: Score improvement wrt Synop
Surface pressure, 24 hr range
Temperature, 24 hr range
Reference
With AMMA RS
Reference
With AMMA RS
Claudia Faccani

Tropical Cyclones in the Indian Ocean: use of Rainy SSM/I data
• Use ECMWF 1D-Var to build a regression between Tb and TCWV on a learning period
•
During experiment, use pre-computed regression to get TCWV pseudo-obs from SSM/I data
•
The obtained TCWV is then assimilated in the ALADIN Reunion model (10km resolution, covers the SWIO)
• A 3D wind bogus following the UKMO technique (Heming, 1995) is used for cyclonic cases
•
Several experiments were ran with the combination of TCWV and the 3D wind bogus
Montroty et al, 2008
Rémi Montroty

Rémi Montroty
Twice as many SSM/I data used

Season statistics in terms of Track Error
TC season 2007, 2007/02/12 to 2007/03/17
Nb of days
UKMO
ECMWF
GFDN
ARPEGE
BOGUS_TCWV
BOGUS
Rémi Montroty Leadtime (h)

Analysis Intensity for TC Indlala
Rémi Montroty

The Concordiasi Experiment over Antarctica
Collaborating institutes
NSF, NCAR, U. Wyoming, Purdue U., UMBC/GMAO,UCLA
CNES, IPEV, CNRS, LGGE, LMD, Météo-France
PNRA
ECMWF
Bureau of Meteorology Research Centre
Concordiasi within International Polar Year
USA
France
Italy
International
Australia
Belongs to the THORPEX-IPY cluster (N
°
121 in IPY)
« Improved numerical weather forecasting and climate simulations by exploitation of in-situ, airborne remote-sensing and satellite data, advanced modelling systems and basic research into polar processes and into polarglobal interactions.
»
Florence Rabier

Field experiment starting in Sept-Nov 2008
 150 radiosoundings from Concordia,
 75 from Dumont d’Urville
 In situ measurements at Concordia
 Stratospheric balloons (probably in 2009)
– Meteorological sensors, ozone sensors
–
Particle counter to study stratospheric clouds
– GPS radio-occultations
 12 driftsondes with 50 dropsondes in each http://www.cnrm.meteo.fr/concordiasi/

Trajectories for late winter/ early spring
(Austral)
NWP users encouraged to use the data, available on the GTS

Operational developments
Positive impact of GPS-RO and IASI data
Large impact of radiance Variational Bias Correction
Encouraging impact of using microwave radiances over land
In Research mode
RS network over Africa in the AMMA context
Use of Cloudy/Rainy radiances for Tropical Cyclones
Field experiment over Antarctica in 2008-2009