Météo-France status

Operational changes since

the 22th North America /

Europe Data Exchange

meeting and short term

plans

Jean-François MAHFOUF

(and many colleagues)

2

Outline

1. Computing platform

2. Model configurations

3. Use of observations

4. Recent operational changes

5. Ongoing developments and future plans

3

Computing platform

Two identical clusters :o Kumo : operationso Yuki : research (operations if

Kumo fails)Each cluster :

o 10 nodeso 16.3 Tflops maxo Interconnected using a

crossbar network (IXS)o 58 Tbytes of scratch (Global

File System)On each node :

o 16 processorso 1 Tbyte of memoryo 2 Tbytes of local disks

Scalar front-end TX7 : 16 Intel Itanium2 cores with 32 Gb memory

Operational configuration

NEC SX9 (Q4 2009)

4

Model configurations

5

Global model ARPEGE (1)

] Spectral model with variable resolution : TL798C2.4L70

o Resolution from 10 km to 60 km, 70 levels from 17m to 0.05 hPao 16 processors for ARPEGE forecast (7’ for 24h forecast)

Forecast ranges [cut-offs] :

FC+102 (00 UTC) [2h15], FC+72 (06 UTC) [3h], FC+84 (12 UTC) [1h50], FC+60 (18 UTC) [3h]

6

Global model ARPEGE (2)

4D-Var assimilation (6h window) :

o 2 loops of minimization : T107C1L70 (25 iterations) + T323C1L70 (25 iterations) – 2nd inner loop with simplified physics (including large scale condensation)

o Variational bias correction scheme since 2008

o Background error variances from an Ensemble Data Assimilation system (4D-Var at lower resolution) since 2008

o 32 processors for assimilation (32’ between cut-off and P0)

Data used :

o SYNOP, SHIP, BUOY, AIREP, AMDAR, ACARS, TEMP, PILOT

o AMV GOES + Meteosat + MTSAT-1R, MODIS (Terra, Aqua), AVHRR/NOAA

o HIRS, AMSU-A, AMSU-B/MHS, NOAA 15, 16, 17, 18, 19 and Aqua

o SSM/IS DMSP F16, 17, 18, AIRS/AQUA, IASI/MetOp, GPS-RO, GPS-ZTD

o Sea surface winds from ASCAT (and ERS2)

o Meteosat CSR

o SST from OSTIA + LST from ½° NCEP + SSM/I F15 for sea-ice mask

7

Evolution of the number of observations in ARPEGE

SCATATOVS

IASI

SEVIRI

SSMI

AIRS

GPS-ROGPS-RO

SEVIRI

AIRS

2002 20062004 2007 2008 2010

8

Information content of observations

9

Summary of ATOVS radiance usage

Satellite HIRS AMSU-A AMSU-B/MHS AIRS/IASI

NOAA15

NOAA16

NOAA17

NOAA18

NOAA19

AQUA

MetOp

: not relevant

: not available

: received but blacklisted

: used

10

Regional model ALADIN

Spectral limited area model : E199x199L70 70 levels from 17m to 0.05 hPa, horizontal resolution 7.5 km

3D-Var assimilation (6h window) : Same data as ARPEGE plus SEVIRI radiances (5 channels over 12)

Current operational domains :

Antilles-Guyana pre-operational

Aladin France

French Polynesia pre-operational

New-Caledonia pre-operational

Aladin La Réunion since 2006

Forecast ranges and cut-offs:

FC+54 (00 UTC) [2h15], FC+48 (06 UTC) [3h], FC+42 (12 UTC) [1h50], FC+36 (18 UTC) [3h]

11

Regional model AROME

Spectral limited area non-hydrostatic model : E359x374L60

o 60 levels from 17m to 0.05 hPa, horizontal resolution 2.5 km

3D-Var assimilation (3h window) :

o Same data as ALADIN plus radar radial winds and reflectivities

Coupling files : hourly ARPEGE forecasts

Forecast ranges and cut-offs :

FC+30 (00 UTC) [1h30], FC+30 (06 UTC) [3h30],

FC+30 (12 UTC) [3h30], FC+30 (18 UTC) [3h30]

12

Additional assimilation and forecasting suites

1. Operational suites for atmospheric models Global stretched with very short cut off [1h10] with 3DVar FGAT (at 00UTC only)

Short-range ensemble prediction system with ARPEGE (TL538C2.4L65) : 35 members. Initial perturbations : singular vectors + EnDA – Model perturbations : random use of 10 different physical packages

Limited area model ALADIN : several research, commercial and transportable dynamical adaptation versions

2. Chemical transport model MOCAGE : air quality forecasts up to 96 h 3 domains : Global – Europe – France with horizontal resolutions of 4°, 0.5°, and

0.1°

Satellite observations currently used only for validation3. Ocean wave model : forecasts up to 102 h

Global, Tropics, Europe, France with horizontal resolutions of 1°, 1°, 0.25°, 0.1°

Assimilation of JASON and ENVISAT altimeter wave height data

13

News on upper-air observations (in 2009)

22 RS

14

News on upper air observations (in 2011)

19 RS

15

Recent operational changes

16

April 2010

Increase in model resolutions : • 15 km -> 10 km for ARPEGE; • 9.5 km -> 7.5 km for ALADIN; • 60 -> 70 levels for ARPEGE/ALADIN; • 41 -> 60 levels for AROME

Increased usage of all satellite observations :o Thinning from 250 to 125 km => 3.5 more data assimilatedo ATOVS data from NOAA 19o Assimilation of low peaking channels from AMSU-B over continents (2)o Assimilation of upper tropospheric wv channels from IASI (9)

Assimilation of radar reflectivities (French network ARAMIS) in AROME

Revised Ensemble Data Assimilation : o 4D-Var instead of 3D-Var FGAT o Use of background error variances for unbalanced temperature, surface

pressure and divergence on top of vorticity.

17

Pre-selection: Only data from detector #1

Pattern depending on scanline

Geographical thinning: 1 prof. / 125km

IASI pixel and channel selection

9 WV

LW - T68 over sea50 over land

channels

18

Thanks to a new surface emissivity model

Example: AMSU-B (atmospheric humidity sounding): only 2 of 5 channels assimilated over land surfaces

Microwave sounding channels from AMSU-B

19

Thanks to a new surface emissivity model

Example: AMSU-B (atmospheric humidity sounding): only 2 of 5 channels assimilated over land surfaces

Microwave sounding channels from AMSU-B

20

November 2010 [1]

Thinning of satellite observations from 125 to 80 km for AROME Assimilation of window channels from SSM/IS F16 and F17 Assimilation of GPS-RO from GRAS-MetOp + for all GPS-RO use

of data between 25 – 36 km Assimilation of low level channels from AMSU-A and AMSU-B

over sea-ice (revised sea-ice emissivity model) Assimilation of more IASI channels and more AVHRR AMVs over

polar regions Improved algorithm for the choice of 2 ambiguous ASCAT winds

among 4

21

CTL EXP

Usage of AMSU-B channels (ch 5: 183.31 7.0 GHz) in ARPEGE

a)a) Assimilation of observations over sea-iceAssimilation of observations over sea-iceOperational since Nov 2010Operational since Nov 2010

Microwave sounding channels from AMSU

22

ScoresScores: Positive in particular over NH: Positive in particular over NH

Wind scores against radiosoundings (one month, January 2009)Wind scores against radiosoundings (one month, January 2009)

Fit to observationsFit to observations: neutral or improvement (temp, profilers, …) : neutral or improvement (temp, profilers, …) Increase about 30% of assimilated AMSU data (sea-ice)Increase about 30% of assimilated AMSU data (sea-ice)

a)a) Assimilation of observations over sea-iceAssimilation of observations over sea-iceOperational since Nov 2010Operational since Nov 2010

Microwave sounding channels from AMSU

23

November 2010 [2]

Increase in AROME domain size (by a factor 1.8) with dedicated

surface analysis OI SST analysis with OSTIA for climatological relaxation Assimilation of SYNOP relative humidity in ARPEGE (daytime

only) Improved procedure for updating anemometer heights on SHIPs

(E-SURFMAR database) EnDA : background error variances for specific humidity + use for

background check Humidity bias correction for RS data following ECMWF Use of saturation pressure formula w.r.t. to water for TEMP

24

Forecast scores : rms Z 500 at 72 hours (00 UTC)

Europe

Northern Hemisphere

25

Current operational changes

E-suite 2011

Inclusion of ATOVS/RARS data from Asia/Pacific + Miami and La

Réunion HRPT stations from EARS

Inclusion of window channels from SSM/IS F18

Inclusion of model error in EnDA

Use of SURFEX (new vegetation/soil climatology + improved ISBA

scheme) in ALADIN/France model

Monitoring of SCAT surface winds from Oceansat2

26

Assimilation of SSM/IS radiances (current e-suite)

00 UTC

12 UTC

27

ARPEGE with very short cut-off (1h10)

AMSU-A from

Exeter (30 %)

EARS AMSU-A

(+ 3 %)

RARS AMSU-A

(+20 %)

Long cut-off

28

Future plans

Preparation of future satellite missionsIncrease (and better) usage of observations currently available

Megha-tropiques : MADRAS + SAPHIR (launch ~ end 2011) NPP : CrIS + ATMS (launch ~ end 2011) IASI on MetOp-B Increase usage of IASI radiances at mesoscale – horizontal error/channel

correlations – improved observation operator Increase usage of IR radiances over land : SEVIRI (S. Guedj thesis) +

IASI (A. Vincensini thesis) Use of IASI cloudy radiances (P. Martinet thesis) ASCAT soil moisture (received operationally since Feb 2011) Preparation of IRS on MTG (Eumetsat fellow) Preparation of wind lidar mission (ESA ADM-Aeolus) GPS data from E-GVAP (Eumetnet) Radar data from OPERA (Eumetnet)

31

Observation usage summary

Datatype Contact Operations Tests

ATOVS elisabeth.gerard@meteo.fr

N15,16,17,18,19,Aqua Metop

AP-RARS

SSM/IS elisabeth.gerard@meteo.fr

F16, F17 F18

Geosat. winds christophe.payan@meteo.fr

GOES, Meteosat, MTSAT

Polar winds christophe.payan@meteo.fr

AVHRR/NOAAMODIS/Aqua, Terra

AVHRR/Metop

Scatterometer christophe.payan@meteo.fr

ASCAT OSCAT

Geosat radiances patrick.moll@meteo.fr

Meteosat GOES, MTSAT

GPS patrick.moll@meteo.frnathalie.saint-ramond

@meteo.fr

ZTDRO

More stationsTerraSar-X/

SAC-CAIRS/IASI vincent.guidard

@meteo.frAqua, Metop More channels

Thank you foryour attention