geneva, may 19-21 st 20084 th wmo workshop on the impact of various observing systems on nwp an...

Geneva, May 19-21st 2008 4th WMO Workshop on the Impact of Various Observing Systems on NWP

An overview of observation impact An overview of observation impact studies performed in the ALADIN studies performed in the ALADIN

communitycommunityClaude Fischer;Claude Fischer;

Zahra Sahlaoui, Fatima Hdiddou (DMN/Maroc);Zahra Sahlaoui, Fatima Hdiddou (DMN/Maroc);Roger Randriamampianina, Gergely Bölöni (Hun. Met. Serv.); Roger Randriamampianina, Gergely Bölöni (Hun. Met. Serv.);

Alena Trojáková (CHMI/Cz. Rep.); Marian Jurašek (SHMU/Slovakia);Alena Trojáková (CHMI/Cz. Rep.); Marian Jurašek (SHMU/Slovakia);Ludovic Auger, François Bouttier, Olivier Caumont, Véronique Ludovic Auger, François Bouttier, Olivier Caumont, Véronique

Ducrocq, Claudia Faccani, Thibaut Montmerle, Eric Wattrelot (Météo-Ducrocq, Claudia Faccani, Thibaut Montmerle, Eric Wattrelot (Météo-France);France);

Günther Haase (SMHI/Sweden)Günther Haase (SMHI/Sweden)


Overview Overview

1.1. Some basic facts on ALADIN and AROMESome basic facts on ALADIN and AROME

2.2. Satellite observations for regional VAR: Satellite observations for regional VAR: SEVIRISEVIRI

3.3. EUCOS/PB-OBS experimentationEUCOS/PB-OBS experimentation

4.4. Radar data:Radar data:1.1. WindWind

2.2. reflectivitiesreflectivities

5.5. Sensitivity of « OSE’s » to other VAR Sensitivity of « OSE’s » to other VAR ingredientsingredients

6.6. Outlook Outlook


Some basic facts on the Aladin and Arome Some basic facts on the Aladin and Arome modelsmodels

► 3D-VAR (no FGAT)3D-VAR (no FGAT)► Continuous assimilation cycles: Continuous assimilation cycles:

Aladin (6 hourly assim, coupled with Arpège, Aladin (6 hourly assim, coupled with Arpège, x=9.5km),x=9.5km), Arome (3 hourly assim, coupled with Ald-Fra, includes NH dynamics and Arome (3 hourly assim, coupled with Ald-Fra, includes NH dynamics and

sophisticated sophisticated ,, x=2.5km)x=2.5km)► Observations:Observations:

Synop: surface pressure,Synop: surface pressure, T2m and RH2m (day), 10m windsT2m and RH2m (day), 10m winds SHIP winds, drifting buoysSHIP winds, drifting buoys Aircraft dataAircraft data SATOB AMV windsSATOB AMV winds Soundings (TEMP RS, PILOT, wind profilers)Soundings (TEMP RS, PILOT, wind profilers) Satellite radiances: NOAA and METOP (AMSU-A/B, MHS, HIRS), Satellite radiances: NOAA and METOP (AMSU-A/B, MHS, HIRS), Meteosat-9 SEVIRIMeteosat-9 SEVIRI scatterometer windsscatterometer winds Ground-based GPS zenital delaysGround-based GPS zenital delays Radar radial winds, radar reflectivity (via RH retrievals)Radar radial winds, radar reflectivity (via RH retrievals)


ALADIN and AROME-France domainsALADIN and AROME-France domains

ALADIN-FRANCEAROME-FRANCE


Satellite radiances for Satellite radiances for regional variational data regional variational data

assimilationassimilation► Impact study with ATOVS/AMSU-A in Impact study with ATOVS/AMSU-A in

Aladin-NORAF (Morocco)Aladin-NORAF (Morocco)► Impact of SEVIRI raw radiances in Impact of SEVIRI raw radiances in

Aladin-FranceAladin-France► Impact of SEVIRI raw radiances in Impact of SEVIRI raw radiances in

Aladin-HungaryAladin-Hungary


TEMPGPS

ATOVS AMSU-B SEVIRI

Seviri data:• WV 6.2 and 7.3 used in clear sky and over low clouds, as classified by the SAF-NWC cloud type product• IR 8.7, 10.8 and 12.0 considered over the sea in cloud free regions only

SEVIRI in Ald-Fra: data coverage as compared SEVIRI in Ald-Fra: data coverage as compared with other obs typeswith other obs types


Sensitivity of the analysis (DFS) per obstypes

OPER: about equal info. Content for TEMP / aircraft / SYNOP & SEVIRI

large increase of DFS for ATOVS data (especially AMSUB) for moreATOVS coupled with a decrease for SEVIRI: the influence of SEVIRI data in the analysis is reduced by the new information brought by extra ATOVS data

without SEVIRI data, DFS values for HIRS and AMSUB almost double

OSEs:

OPER: ALADIN/France oper

moreATOVS: OPER with denser ATOVS data (AMSUA, AMSUB and HIRS) : 1 pixel/80 km vs. 1/250 km)

noSEV: OPER without SEVIRI

Montmerle etal,, QJRMS, 2007

Impact of radiances from polar orbiting v/s Impact of radiances from polar orbiting v/s geostationary sat.geostationary sat.

)(y

HxTrDFS

a


Impact of SEVIRI radiances in Ald-HungaryImpact of SEVIRI radiances in Ald-Hungary

►RMSE 90% RMSE 90% significance significance testtest

►Top: conv. Top: conv. Data + SEVIRIData + SEVIRI

►Bottom: conv. Bottom: conv. Data + SEVIRI Data + SEVIRI + SYNOP 2m T + SYNOP 2m T & RH& RH


Impact of SEVIRI radiances in Ald-Hun: RR for Impact of SEVIRI radiances in Ald-Hun: RR for all thresholds and all fcst lead timesall thresholds and all fcst lead times


Roger RANDRIAMAMPIANINARoger RANDRIAMAMPIANINAHungarian Meteorological Service, Budapest, HungaryHungarian Meteorological Service, Budapest, Hungary

[email protected]@met.huContributorsContributors::-- Gergely Gergely BÖLÖNI, Sándor KERTÉSZBÖLÖNI, Sándor KERTÉSZ, , Andrea LŐRINCZ and András Andrea LŐRINCZ and András HORÁNYIHORÁNYI

Space-Terrestrial StudySpace-Terrestrial StudyEUCOS/EUCOS/PB-PB-OBS seminarOBS seminar

Domain of Domain of the the ALADIN/HU ALADIN/HU model:model:

12km, 37 levels, 48 h forecasts 12km, 37 levels, 48 h forecasts twice a daytwice a day


- Active observations- Active observations

- Rejected - Rejected observationsobservations

Example of QC Example of QC (first guess (first guess

check)check)


Winter/Summer Winter/Summer

EUEU01/E01/ESS01- baseline (GSN surface and GUAN radiosonde + AMV + 01- baseline (GSN surface and GUAN radiosonde + AMV + ATOVS rad.)ATOVS rad.)EU02/EEU02/ESS02- baseline + 02- baseline + aircraftaircraft EU03/EEU03/ESS03- baseline + 03- baseline + radiosonde windradiosonde windEU04/EEU04/ESS04- baseline + radiosonde wind and 04- baseline + radiosonde wind and temperaturetemperatureEU05/EEU05/ESS05- baseline + 05- baseline + wind profilerswind profilersEU06/EEU06/ESS06- baseline + 06- baseline + radiosonde wind and temperatureradiosonde wind and temperature + + aircraftaircraft EU07/EEU07/ESS07- baseline + radiosonde wind , temperature and 07- baseline + radiosonde wind , temperature and humidityhumidityEU08/EEU08/ESS08- full observation (radiosonde + aircraft + wind profiler)08- full observation (radiosonde + aircraft + wind profiler)

Description of the Description of the experimentsexperiments


Forecast evaluation: Impact Forecast evaluation: Impact of of radiosonde wind radiosonde wind observationobservation

Temperature T-850Temperature T-850

Forecast evaluation: Impact Forecast evaluation: Impact of of radiosonde temperature radiosonde temperature observationobservation

Temperature T-850Temperature T-850


Forecast evaluation: Impact of the radiosonde Forecast evaluation: Impact of the radiosonde humidityhumidity

00 UTC 12 00 UTC 12 UTCUTC

BLBL + radiosonde T,V + radiosonde T,V BL + BL +radiosonde T,V,HU radiosonde T,V,HU Full obs. Full obs.RMSERMSE

RR (mm/6h): comparison against surface measurements RR (mm/6h): comparison against surface measurements


Overview of the most striking impacts:Overview of the most striking impacts:► Radiosonde:Radiosonde:

Wind: positive impact in the troposphere over day 1Wind: positive impact in the troposphere over day 1 T: positive impact over day 1; strong positive signal on MSLP for the T: positive impact over day 1; strong positive signal on MSLP for the

summer periodsummer period RH: clear positive impact on MSLP in the winter period; large positive RH: clear positive impact on MSLP in the winter period; large positive

impact on RR at all fcst rangesimpact on RR at all fcst ranges► Aircraft AMDAR:Aircraft AMDAR:

W/R to ECMWF analysis: clear positive signal on T, W/R to ECMWF analysis: clear positive signal on T, , FF, RH at all fcst , FF, RH at all fcst rangesranges

Bigger impact in the summer period on all fields, up to 24 hBigger impact in the summer period on all fields, up to 24 h Positive impact on RH fcst in the summer, but negative impact for the Positive impact on RH fcst in the summer, but negative impact for the

winter periodwinter period Positive impact on RR fcst, mostly in the summerPositive impact on RR fcst, mostly in the summer

► AMDAR T/wind on top of RS T/wind:AMDAR T/wind on top of RS T/wind: Larger (positive) impact for the summer period, with an AMDAR extra Larger (positive) impact for the summer period, with an AMDAR extra

effect about half of the one of RSeffect about half of the one of RS


Radar data assimilationRadar data assimilation

►Doppler radial winds (Vr)Doppler radial winds (Vr)►Reflectivities Reflectivities


MSLP

T2m RH2m

FF

DD

RR

R0

CNTRL RADAR

Vr: neutral to Vr: neutral to slightly positive slightly positive

scores in AROME-scores in AROME-FranceFrance


RADAR

CNTRL

RADAR

CNTRL

Guess

01/12/0718 UTC

Analysis of Analysis of divergence at 950 divergence at 950

hPahPa


Goal Goal ::Operationnally assimilate radar reflectivities in AROME by 2009-Operationnally assimilate radar reflectivities in AROME by 2009-

20102010

Method :Method :► Volumic (3D) reflectivity data routinelyVolumic (3D) reflectivity data routinely available at MF since available at MF since

August 2007, in real time. Pre-processing check to remove August 2007, in real time. Pre-processing check to remove erroneous data (soil and sea clutters, …)erroneous data (soil and sea clutters, …)

► Reflectivity observation operator readyReflectivity observation operator ready, simulates modelled , simulates modelled reflectivities.reflectivities.

► Quality control checkQuality control check by a gross comparison of observed and by a gross comparison of observed and modelled columns.modelled columns.

► Assimilation in the AROME system via a 1D+3DVarAssimilation in the AROME system via a 1D+3DVar: : reflectivities are inverted into pseudo-observations of relative reflectivities are inverted into pseudo-observations of relative humidity profiles (whose impact is expected to be bigger than humidity profiles (whose impact is expected to be bigger than when modifying the hydrometeor fields).when modifying the hydrometeor fields).

Radar reflectivity assimilationRadar reflectivity assimilation


Sensitivity of « OSE’s » to B Sensitivity of « OSE’s » to B matrixmatrix

MSLP 48h

850 hPa wind 0h

850 hPa wind 12h

Change of B matrix file: from theinterpolated 46/60 to the genuine60 level ensemble fct stats (Jan. 19th)

Impact of using an interpolated B (here, vertical levels changed from 46 to 60) in the E-suite,W/r to using the genuine 60 levelRecomputed ensemble fct stats(work of Abdelhak Razagui, ALG)


Outlook Outlook ► Future OSE’s: Arome on COPS period (South-West of Future OSE’s: Arome on COPS period (South-West of

Germany, July/August 2007)Germany, July/August 2007)► New observations for LAM assimilations:New observations for LAM assimilations:

Microwave radiances over landMicrowave radiances over land Radar data: 2008 (radial winds), 2009 or 2010 (reflectivities)Radar data: 2008 (radial winds), 2009 or 2010 (reflectivities) Cloud boguses ?Cloud boguses ? Combine upper-air analyses with a surface assimilation (simplified Combine upper-air analyses with a surface assimilation (simplified

EKF)EKF)► Other R&D aspectsOther R&D aspects

For optimized obs impact: For optimized obs impact: one needs to work on all aspects of the one needs to work on all aspects of the assimilation systemassimilation system: B model, 4D-VAR ?, ensemble system : B model, 4D-VAR ?, ensemble system simulation, filtering and spin-up, …simulation, filtering and spin-up, …

Common Aladin-Hirlam 4-year plan Common Aladin-Hirlam 4-year plan

Thank you for your Thank you for your attentionattention


Impact of AMSU-A is positive over areas with a lack of observations Impact of AMSU-A is positive over areas with a lack of observations (tropical band, Sahara, ocean)(tropical band, Sahara, ocean)

In mid latitudes, the impact of AMSU-A raw radiances is less marked In mid latitudes, the impact of AMSU-A raw radiances is less marked (conflict between several observation types). More sensitivity (conflict between several observation types). More sensitivity studies are needed to optimise the use of raw radiancesstudies are needed to optimise the use of raw radiances

P30 humidity field at 850hPa level

Impact study of NOAA/ATOVS raw radiances in Impact study of NOAA/ATOVS raw radiances in Aladin/NORAF (Morocco)Aladin/NORAF (Morocco)


For moreATOVS, DFS/p is slightly reduced for HIRS and AMSUB : using more data decreases the individual influence of one radiance in the analysis

This reduction is accentuated for AMSUA because of the broad structure functions that are used in the high troposphere/low stratosphere

For noSEV, the individual influence of one radiance is much higher than in OPER : This shows the complementarity of datasets that are sensitive to the same atmospheric component (very interesting for r6 and r18)

Montmerle et al, 2006

(DFS/nb of obs) ratio per obs type(DFS/nb of obs) ratio per obs type


- Comparison against ECMWF analyses: a Comparison against ECMWF analyses: a clear positive impactclear positive impact on the temperature, geopotential, wind speed and humidity on the temperature, geopotential, wind speed and humidity fields was foundfields was foundfor all the forecast ranges for all the forecast ranges

- Comparison against observations: the impact concerned - Comparison against observations: the impact concerned mostly mostly the analysis and forecasts up to 24-hour the analysis and forecasts up to 24-hour

- A - A bigger positive impact of the aircraftbigger positive impact of the aircraft observation was observation was observed for the summer period than for the winter periodobserved for the summer period than for the winter period

- - Positive impact of the aircraftPositive impact of the aircraft data on the forecast of humidity data on the forecast of humidity fields was observed for the summer period, while fields was observed for the summer period, while negative negative impact impact was found for the winter periodwas found for the winter period

- - Positive impact of the aircraftPositive impact of the aircraft data on the forecast of data on the forecast of precipitation was observed for the summer period, while precipitation was observed for the summer period, while neutral neutral (from 00UTC) and negative (from 12 UTC) impact was(from 00UTC) and negative (from 12 UTC) impact was observed observed for the winter period for the winter period

Impact of aircraft (AMDAR) observationsImpact of aircraft (AMDAR) observations


-In the troposphere, In the troposphere, clear positive impactclear positive impact of the of the radiosonde windradiosonde wind observation on the analysis and short-range forecasts was observation on the analysis and short-range forecasts was observed observed

- A - A positive impactpositive impact of the of the radiosonde temperatureradiosonde temperature on the analysis on the analysis and on the forecasts up to (mostly) 24-hours was observed and on the forecasts up to (mostly) 24-hours was observed

- - Clear positive impact Clear positive impact of theof the radiosonde temperature radiosonde temperature data on the data on the analysis and on the forecasts of the analysis and on the forecasts of the mean sea level pressuremean sea level pressure up to up to 24-hours was observed for the summer, while 24-hours was observed for the summer, while neutral impactneutral impact was was found for the winter periodfound for the winter period

- - Neutral impact Neutral impact of the of the radiosonde humidityradiosonde humidity on the on the mean sea level mean sea level pressurepressure was observed for the summer period, while was observed for the summer period, while clear positive clear positive impactimpact was observe was observedd for the winter period for the winter period

- - Bigger positive impactBigger positive impact of the of the radiosonde temperature radiosonde temperature on the on the geopotential was observed in the summer study than for the winter geopotential was observed in the summer study than for the winter study study

- - Large positive impact Large positive impact of theof the radiosonde humidity radiosonde humidity was observed was observed for all forecast ranges of precipitation for all forecast ranges of precipitation

Impact of radiosonde observationsImpact of radiosonde observations


- Comparing ES01, ES02 and ES04 (summer study), we found that - Comparing ES01, ES02 and ES04 (summer study), we found that the impact of the the impact of the aircraft (wind & temp.)aircraft (wind & temp.) observations was a bit observations was a bit larger than what we found during the winter study (- half of that larger than what we found during the winter study (- half of that of the of the radiosonde (wind & temp.)radiosonde (wind & temp.) data) data)

- For the summer period, a small improvement in the scores was - For the summer period, a small improvement in the scores was observed when comparing the impact of the observed when comparing the impact of the aircraft dataaircraft data on top on top of the of the radiosonde windradiosonde wind and temperature dataand temperature data (ES04 vs ES06), (ES04 vs ES06), while a small deterioration was observed in the winter studywhile a small deterioration was observed in the winter study

Impact of AMDAR T/wind on top of RS T/windImpact of AMDAR T/wind on top of RS T/wind


10m wind observations in ALADIN-10m wind observations in ALADIN-FranceFrance

► ~2750 stations from synoptic network~2750 stations from synoptic network► Monitoring over 4 month data (September 2006 – December 2006)Monitoring over 4 month data (September 2006 – December 2006)► Blacklisting when correlation between obs and model value < 0.3 Blacklisting when correlation between obs and model value < 0.3

=> 101 stations blacklisted=> 101 stations blacklisted► Slight improvement with blacklistingSlight improvement with blacklisting► Experimental period : 01/09/2006-15/09/2006Experimental period : 01/09/2006-15/09/2006► Good scores, especially in terms of sea-level pressure, Good scores, especially in terms of sea-level pressure,

tropospheric wind and tropospheric humiditytropospheric wind and tropospheric humidity

Comparison to pressure surface observation Scores to TEMP for wind force

RMS STDEV BIAS


Elimination of non-Gaussian Elimination of non-Gaussian innovations ?innovations ?

► Test with removal of 600 stations (out of 2753) showing Test with removal of 600 stations (out of 2753) showing the less Gaussian aspect for innovationsthe less Gaussian aspect for innovations

► Test based on good amount of population inside the 3 Test based on good amount of population inside the 3 thresholds : thresholds : 2 2 3 3

► No improvement, instead slight deteriorationNo improvement, instead slight deterioration

Deterioration of wind Scores compared with TEMP

Bad station (PDF tooFar from Gaussian profile)

Good station

RMS STDEV BIAS


One needs to go to QPF scores and fine-scale flow analysis (field of horizontal divergence)

Severe wind situation on Northern France (early December 2007):

non Doppler radarsDecember 3rd, 00 UTC

area of interest is well covered with radar wind information

More pertinent assessment of the impact of radar More pertinent assessment of the impact of radar data:data:


Malfunctions at Bollène and St Nizier

QPF scores:A more difficult period for the radar experiment: Nov. 19th, 00 UTC -> Nov. 24th 00 UTCPrecipitations over Eastern France

RADAR

CNTRL

A case with non-operating radarsA case with non-operating radars


Histograms:• Blue: total number of observations• Yellow: number of observations entering minimization

Red: number of elevations

Black: mean variance of Vr computed over slices of 10 deg. Site angle, per elevation

Example of 15 days for radars in Trappes, Abbeville, Falaise and AvesnesMonitoring of radar radial wind observationsMonitoring of radar radial wind observations


Resolution volum, ray path : standard refraction (4/3 Earth’radius)

zh

rd

eZ

r

Nd

d

model level

• Bi-linear interpolation of the simulated hydrometeors (T,q, qr, qs, qg) • Compute « radar reflectivity » on each model level

Backscattering cross section: Rayleigh (attenuation neglected)

..., 0

),().,()(snowrainj

dDrDNjrDjr

Microphysic Scheme in AROME

Diameter of particules

• Simulated Reflectivity factor in « beam volum bv»

Antenna’s radiation pattern: gaussian function for main lobe

(side lobes neglected)

)..).,().(log10 4 dddrfr(Zbv

e

Resolution volum, ray path: standard refraction (4/3 Earth’s radius)

Radar observation Radar observation operator geometry operator geometry implemented in the implemented in the

ALADIN/AROME 3D-VARALADIN/AROME 3D-VAR

geneva, may 19-21 st 20084 th wmo workshop on the impact of various observing systems on nwp an...

Documents

data seviri synop

influence of seviri

data seviribottom

data coverage

extra atovs data

t rhimpact of seviri

denser atovs data amsua

francesatellite radiances