deutscher wetterdienst status report
DESCRIPTION
19th North America-Europe Data Exchange Meeting Silver Spring, Maryland May 3-5, 2006. Deutscher Wetterdienst status report. Reinhold Hess. Silver Spring, 2006. Deutscher Wetterdienst status report. Overview: Events in NWP since the last report - PowerPoint PPT PresentationTRANSCRIPT
Deutscher Wetterdienst status reportDeutscher Wetterdienst status report
Reinhold Hess
Silver Spring, 2006
19th North America-Europe Data Exchange MeetingSilver Spring, Maryland May 3-5, 2006
Reinhold Hess, 2
Deutscher Wetterdienst status report
Overview:• Events in NWP since the last report
• use of satellite radiances with 1D-Var (AMSU-A)• use of AMV of Meteosat 8 (and MTSAT)
• Research• AIRS/IASI with 1D-Var• Scatterometer (QuikSCAT)• Radio Occultation• LME: SEVIRI and ATOVS with 1D-Var• LMK: Radar reflectivities with LHN• 3D-Var
• Conclusion
Silver Spring, 2006
Reinhold Hess, 3
Global Model GME 40/40
Model System:
GME (Global Model): hydrostatic, icosahedral-hexagonal grid, mesh size ~40kmterrain following hyb. coordinate, 40 layers, bottom layer at 10 m, 7 soil layersforecast range: 174h initial dates 00, 12 UTC and 48h for 18 UTCprognostic cloud ice, prognostic sea ice
Analysis: OI, 3-hourly intermittent analysis, observation window +/- 1.5hobservations: conv., AIREP, AMDAR, ACARS, ATOVS (NOAA-15,16 and AQUA),
AMV (GOES E and W, GMS, Meteosat 5 and 7), PAOBMODIS-Winds, PSEUDO-tempscutoff: 2h30 (for forecasts)
Silver Spring, 2006
Reinhold Hess, 4
Assimilation of IFS Pseudo Temps
Data Coverage
Profiles of IFS/4D-Var Analyses (temperature, humidity and wind)as Temps for OI/GME
Assimilation one time per day for 0 UTCinto Main Analysis only
(operational since Dec 17 2003)
use of humidity profiles only above 700 hpa
Silver Spring, 2006
Reinhold Hess, 5
Use of ATOVS with 1D-Var
Setup: • Level 1c ATOVS from UKMO (NOAA 15, 16 and AQUA)
(NOAA 18 in preparation)
• Channel Choice: AMSU-A 5-12 (HIRS deferred)
• Bias-Correction: UKMO-Style (scan angle and mass, A4 and A9)
• MW Rain/Ice-detection (Kelly and Bauer/ECMWF)
• Assimilation as SATEMS in OI (geopotential thicknesses, PWC)
(assimilate 1d-var analysis increments only)
• Retrieval of temperature only (humidity deferred)
• B-Matrix based on GME-Statistics (NMC-Method)
• Use of IFS forecasts above GME model top (10hPa)
Silver Spring, 2006
Reinhold Hess, 6
• Significant improvement of conventional Routine (no Pseodo-Temps) with 1D-Var• 1D-Var (without PT) is comparable to Routine (with PT) for 0 UTC
500 hPa ANOC, 0 UTC, Routine, 1006 (conv. Rou), 1004 (1D-VAR)
Silver Spring, 2006
Reinhold Hess, 7
500 hPa ANOC, 12 UTC, Routine, 1006 (conv. Rou), 1004 (1D-VAR)
• Significant improvement of conventional Routine (no Pseodo-Temps) with 1D-Var• 1D-Var (without PT) is lags behind Routine (with PT) for 12 UTC
Silver Spring, 2006
Reinhold Hess, 8
500 hPA ANOC, 0 UTC, for Routine, 4638 (Pseudo+1D-VAR)
1D-Var (with PT) further improves compared to Routine (with PT) for 0 UTC
Silver Spring, 2006
Reinhold Hess, 9
500 hPA ANOC, 12 UTC, für Routine, 4638 (Pseudo+1D-VAR)
1D-Var (with PT) is neutral compared to Routine (with PT) for 12 UTC
Silver Spring, 2006
Reinhold Hess, 10
Use of ATOVS with 1D-Var
Monitoring with routine GME available on Internet:http://www.dwd.de/en/FundE/Analyse/Assimilation/
> Monitoring
diurnal cycles of o-b differences are caused byassimilation of pseudo-temps one time per daybecause of slightly different inherent model climatesof GME and IFS/ECMWF
Silver Spring, 2006
Reinhold Hess, 11 Silver Spring, 2006
Reinhold Hess, 12
Use of AIRS/IASI
Status:• monitoring available• cloud detection implemented (S. English)• validation of channel selection (UKMO sample)• bias correction (mass and scan angle correction)• channel-based cloud detection implemented (T. McNally)
validation study at ECMWF
Next:• trial experiments (just started)• validation of retrievals and forecasts• preparation for IASI
Silver Spring, 2006
Reinhold Hess, 13
collocation of AIRS-787 departures,MODIS visual cloud product,model clouds, and results of AIRS cloud detection
distributions of AIRS-787 first guess departuresfor visually cloud free and cloudy fovs
Validation study of channel selection at ECMWF
inconsitencies detected and clarified,e.g. visual cirrus does not affect lw CO2 band
visually cloud free sample has bias of 0.04 K,visually cloud affected sample has bias of -0.07K
Silver Spring, 2006
Alexander Cress
• Usage• extratropics over oceans; tropics over oceans and land
• IR above 1000 hPa • QI > 85
• WVcloudy above 400 hPa (both channels)• QI > 80; tropics > 85
• WVcloudy between 700 and 400 hPa (7.3 μm)• QI > 80 only on the Southern Hemisphere
• WVclear is not used• VIS below 700 hPa
• QI > 65 ; tropics: QI > 85
• FG check: asymmetric to remove negative OBS-FG bias
• Thinning: 1 wind per pre-defined thinning box (200 km;15 vertical layers).
Analysis and forecast impact of AMVs of Meteosat 8(generated at EUMETSAT)
Alexander Cress
Difference of the mean RMS increments between an analysis using Meteosat 7 and Meteosat 8 AMV winds
• Reduction of RMS by replacing Meteosat 7 AMV winds with Meteosat 8 wind data• Positive impact of Meteosat 8 AMV winds on the global GME analysis
Alexander Cress
NH
SH
TR
EU
forecast time h forecast time h
anco
ran
cor
Routine
Exp. 5344
Routine Routine
Routine
Exp. 5344 Exp. 5344
Exp. 5344
Routine with Meteosat 7 AMVs
Experiment with Meteosat 8 AMVs
replacing Meteosat 7
Anomalie correlation coefficient for 500 hPa geopotential height date: 2005111500 - 2005121500, 30 forecasts
Use of QuikSCAT DataQuikSCAT data 25 km resolution
date: 2004092709Used data JPL Rain Flag Used data KNMI Rain Flag
Silver Spring, 2006
Reinhold Hess, 18
Use of QuikSCAT Data
Obs
bias: 0.80rms: 3.21cor: 0.66
All DataData with KMNI Rain Flag 0without outer zone and full bias correction
bias: 0.22rms: 2.06cor: 0.82
F G F G
Status:• monitoring available• KNMI rain flag• bias correction• test cases
Next:• more trial runs
Silver Spring, 2006
Obs -FG (All data, bcor) Obs - FG (OI data, bcor)
STDV Obs -FG (All data, bcor) STDV Obs -FG (OI_data, bcor)
Mean: 0.63
Mean: 1.85
Mean: 0.44
Mean: 1.48
• Data selection (KNMI rain flag) reduces Bias and Stdv. • Some rain contamination still present in tropics ?
Statistics for QuikSCAT wind speed Dez. 2004
Reinhold Hess, 20
Use of Radio-Occultation Data (CHAMP)
Receiving Produkt from MPI-Hamburg(bending angles based on canonical transformation)
Monitoring since Jan 2004
bias and stddevof bending anglescompared to GME [%]
Work deferred until COSMIC data becomes availableSilver Spring, 2006
Reinhold Hess, 21
LM (Local Model): non-hydrostatic, rotated latitude-longitude grid, mesh size 7kmterrain following hyb. coordinate, 40 layersforecast range: 78 for initial dates 00, 12prognostic cloud ice, prognostic rain
Model System:
Analysis: continuous nudgingobservations: conv., AIREP,AMDAR,ACARScutoff: 2h30variational soil moisture analysis
Local Model LM (LME)
Silver Spring, 2006
Reinhold Hess, 22
MW rain/ice detection (Kelly/Bauer) as for GME
• Assimilation of 1D-Var retrievals with nudging analysis• Continuous nudging analysis requires 1D-Var within LM for topical background
Use of Radiances (SEVIRI/MSG and ATOVS) for LME with 1D-Var
Silver Spring, 2006
NWC-SAF scene analysis for IR/WV cloud detection
Status: Trial runs about to start
Model Domain LMK
LMK:• x = 2.8 km• 50 levels• t = 30 sec.• Tges = 18 h
– Initialised with LME every 3 hours
– Explicit simulation of deep convection
Silver Spring, 2006
Latent Heat Nudging of Radar Reflectivities for LMK
Silver Spring, 2006
Reinhold Hess, 25
Minimisation in Observation Space rather than Model Space
Conventional 3D-VAR (MSAS, Model Space Assimilation System)solve:
(physical space assimilation system, dual space, observation space)
PSAS (OSAS, Observation Space Assimilation System)solve:
• more flexibility in definition of B• observation space smaller than model space !?!• minimisation costs become quadratic in number of observations
Development of PSAS (3D-Var)
Silver Spring, 2006
Reinhold Hess, 26
Development of PSAS (3D-Var)
Adaptive Vertical Error Correlations for 1D-Var
(depending e.g. on vorticity)
Silver Spring, 2006
Reinhold Hess, 27 Silver Spring, 2006
Reinhold Hess, 28 Silver Spring, 2006
Reinhold Hess, 29
Status and Plans
Conclusion:
• Operational: ATOVS/AMSU-A with 1D-Var and OI (substituting SATEMS)• Pre-operational: AMVs of MET-8 (substituting MET-7)• Operational Plans in
• ATOVS/AMSU-B• AIRS/IASI• QuikSCAT• Radio Occultation
• Prototype of PSAS/3D-Var (direct Assimilation of ATOVS, AIRS)• but also Interest in AIRS/IASI-Retrievals
Silver Spring, 2006
Reinhold Hess, 30
Many thanks foryour attention!
Deutscher Wetterdienst status report
Silver Spring, 2006