data exchange meeting, dorval 26-28 may 04 met office report relocation model developments use of...
TRANSCRIPT
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
Relocation
Met Office HQ relocated ~200km from Bracknell to Exeter - biggest IT move in Europe!
Operational forecasts started in Exeter Sep 03
New building completed in Dec 03
More than 70% of the staff moved to Exeter
All open plan offices (including Director!)
New director will be Dave Rogers
BracknellExeter
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
Met Office NWP ModelsData Assimilation: 3DVar, FGAT, 6 hr cycle1hr 50min cut-off with 7hr update runs for next cycle
Model formulation: Exact equations of motion in 3D, non-hydrostatic effects included, semi-Langrangian scheme, hybrid-eta in height.
Data Exchange Meeting, Dorval 26-28 May 04
Global Model No major upgrades over the last year due to relocation
‘freeze’ Successfully migrated to NEC SX-6 from Cray T3E’s on
27 April 04. Satellite upgrade goes operational this week more later 4DVar planned for late summer 04 implementation,
conservative implementation but good impacts. 70L model with top above 0.1hPa planned for 2005
along with increased horizontal resolution. 16 member ensembles out to T+72 at half operational
resolution planned for 2005.
Data Exchange Meeting, Dorval 26-28 May 04
European model Domain
Current UK mesoscale
New Euro Model 20km resolution
548
320
Data Exchange Meeting, Dorval 26-28 May 04
European Model (EUROLAM)
20km grid model became operational on 23 Mar 04
Run 4 times per day out to T+48hr Data assimilation 3DVar Observation usage similar to global model Products planned to be available in late
2004 Move to 12km grid in 2005
Data Exchange Meeting, Dorval 26-28 May 04
Ob type Notes Global/
Euro UK
Mesoscale Synops Surface Pressure
U,V (land)
U,V (sea)
Temperature
Relative Humidity
Visibility
Drifting buoys Aircraft AIREPS, AMDARS Sondes TEMP, PILOT, Dropsondes Satellite atmospheric motion winds
Meteosat-5/7 (IR,WV, VIS)
Meteosat-7 (IR,WV, VIS)
GOES-10/12 (IR)
GOES-9 (IR)
MODIS (Terra/Aqua)
Scatwinds Locally processed ERS scatterometer
SeaWinds (Quikscat)
ATOVS Radiances Wind profilers European and US wind profilers . AIRS Subset of radiances
SSM/I 10m windspeed
Total column water vapour
MOPS Satellite derived cloud product
Observing systems used or being assessed for use in the data assimilation system. A tick in the 3rd or 4th column indicates operational use.
Observations
used in
Global, Euro
and UK
Mesoscale
models
(May 04)
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
Met Office NOAA Data Acquisition
Local
HRPT EARS
1a data
Global
1b data
WashingtonWest Freugh
or ExeterEUMETCAST
AAPP AAPP AAPP
METO NWP
1d data on HIRS grid
• 1a counts + housekeeping
• 1b counts + navigation + cal info
• 1c brightness temps
• 1d brightness temps on HIRS Grid
1c data to
European Met
Centres
Data Exchange Meeting, Dorval 26-28 May 04
Use of ATOVS data Global level 1b data from NESDIS Converted to level 1d (all IFOVS mapped
to HIRS) using AAPP software Level 1d radiances pre-processed with
1D-Var Level 1d radiances assimilated in 3D-Var HIRS/AMSU in Global and AMSU only in
Euro and mesoscale models Main global model run (cut-off 1:50) Update model run (cut-off ~7:00)
Data Exchange Meeting, Dorval 26-28 May 04
Use of EARS ATOVS data Receiving ATOVS 1A data from up to 8 stations Routinely compare level 1B global (from NESDIS)
with EARS level 1B (using AAPP to convert from 1A to 1B)
Comparisons of:
– Earth location
– Brightness temperatures HIRS, AMSU AMSU data show good agreement but HIRS shows
large differences due to calibration To reduce the delay in the ATOVS ‘blind’ orbits the
EUMETSAT EARS AMSU data are now being assimilated in our global forecast model
Data Exchange Meeting, Dorval 26-28 May 04
EARS data coverage
STATIONS
Black: Current
Red: Proposed future upgrade
Data Exchange Meeting, Dorval 26-28 May 04
Arrival Times of Data
• North Atlantic Region
• six-hour window
09/09/2003 09:00-15:00
window
0
10
20
30
40
50
60
70
80
90
100
09 00 10 00 11 00 12 00 13 00 14 00 15 00 16 00 17 00 18 00
Pe
rce
nta
ge
of
Ob
se
rva
tio
ns
Global 109229 obs EARS 93979 Obs
Main RunNESDIS 25%EARS 75%
Update Run both 100%
First Overpass
Data Exchange Meeting, Dorval 26-28 May 04
Experiment to assess impact of early cut-off on
forecasts Period 12 May – 4 June 2003 NOAA level 1b radiances used:
– NOAA-15 AMSU-A/AMSU-B
– NOAA-16 HIRS/AMSU-A/AMSU-B
– NOAA-17 HIRS/AMSU-A/AMSU-B
Control: All data received before 1:50 (Ops) for main run and all late ATOVS used in update run
Experiment: All data received before 1:50 + all late ATOVS received (up to 50% more data) used in main model run
Data Exchange Meeting, Dorval 26-28 May 04
All ATOVS forecast impacts
Effect of including ALL ATOVS
-7
-6
-5
-4
-3
-2
-1
0
1
2
T+24 T+48 T+72 T+96 T+120 T+24 T+48 T+72 T+24 T+24 T+48 T+72 T+24 T+24 T+48 T+72 T+96 T+120 T+24 T+48 T+72 T+24
PMSL PMSL PMSL PMSL PMSL H500 H500 H500 W250 W850 W850 W850 W250 PMSL PMSL PMSL PMSL PMSL H500 H500 H500 W250
NH NH NH NH NH NH NH NH NH TROP TROP TROP TROP SH SH SH SH SH SH SH SH SH
Parameter
RM
S F
C e
rro
r %
dif
f
Verified by Obs Verified by Analysis
Data Exchange Meeting, Dorval 26-28 May 04
Forecast impacts
N. Hem500hPa
1:50 cut-off
Improved bust?
Data Exchange Meeting, Dorval 26-28 May 04
Forecast impacts
S. Hem500hPa
1:50 cut-off
Consistent improvement
Data Exchange Meeting, Dorval 26-28 May 04
Example985 hPa low in control
975 hPa low including late ATOVS
976 hPa low in analysis
Data Exchange Meeting, Dorval 26-28 May 04
EARS forecast impactsUpper Level Windspeed (250 hPa)
5
7
9
11
13
15
17
19
21
24 48 72 96 120 144
Forecast Range (hours)
RM
S E
rro
r (m
s-1)
EARS+Global
Global
Data Exchange Meeting, Dorval 26-28 May 04
AMSU ‘precip’ imagery
increasing cloud liquid water incr. scattering
Imagery Application
Other applications outside of NWP can also exploit the timeliness of EARS data. This plot shows an example of a now-casting product that is disseminated in near real time to forecasters. The areas coloured green, red, yellow denote increasing cloud liquid water, identified using AMSUA radiances. Blue denotes the scattering signal in the AMSUB radiances, produced either by ice particles in the cloud, which can infer precipitation, or sea-ice. The product is overlaid on a coincident Meteosat IR image.
Data Exchange Meeting, Dorval 26-28 May 04
Summary of recent OSEs at the Met Office
Remove AMSU, HIRS, AMSU+HIRS radiances
Remove radiosonde temperature and humidity
Remove 1, 2, 3 x ATOVS CHAMP Radio Occultation
Data Exchange Meeting, Dorval 26-28 May 04
Percentage change in RMS fit of radiosonde Z & RH, and
ship/buoy/synop pmsl to T+6
-14
-12
-10
-8
-6
-4
-2
0
2
Per
cen
tag
e ch
ang
e in
RM
SE
Z2
50
Z5
00
Z7
00
Z8
50
No AMSU No HIRS No ATOVS No Sonde T and RH
Data Exchange Meeting, Dorval 26-28 May 04
Schematic
No AMSU and no HIRS
No AMSU so HIRS only
No HIRS AMSU onlyAMSU and HIRS
~6%~6%
~2%
~4%
~ 0%
So HIRS >> nothing. So assimilating only in cloud-free areas does improve forecasts. So AIRS should give a benefit in data sparse areas? Worth using HIRS if AMSU-A fails (as on NOAA-17)?
Data Exchange Meeting, Dorval 26-28 May 04
Percentage change in RMS fit of radiosonde Z & RH, and
ship/buoy/synop pmsl to background
-14
-12
-10
-8
-6
-4
-2
0
2
Per
cen
tag
e ch
ang
e in
RM
SE
Z2
50
Z5
00
Z7
00
Z8
50
Sh
ip &
bu
oy
pm
sl
Sy
no
pp
ms
l
3-> 0 ATOVS 3-> 1 ATOVS 3-> 2 ATOVS
Data Exchange Meeting, Dorval 26-28 May 04
Conclusions Late ATOVS missing main f/c run degrades forecasts. EUMETSAT EARS can help alleviate this. First and second AMSU are very important to NWP. Third AMSU still positive overall, but main role is
robustness. HIRS impact much less than AMSU but still an impact
on low level moisture (NB SSM/I low level moisture not used). In absence of AMSU, HIRS impact is still very significant so important for robustness.
AIRS initial impacts much larger than HIRS, smaller than sonde temperatures and much smaller than AMSU.
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
Monitoring web page
Available via password protected pageon Met Office site.
Userid:airspagePasswd: &Graces
Data Exchange Meeting, Dorval 26-28 May 04
Change in Forecast Errors:500hPa Height at 24 hours
-0.2%
-0.6%
-1.8%
Red/green reduced errors due to AIRS
Data Exchange Meeting, Dorval 26-28 May 04
Satellite Spring UpgradeAndrew Collard, Nigel Atkinson, Brett Candy, Fiona Hilton, James
Cameron, Roger Saunders and Stephen English
AIRS radiances: cloud-free, ocean fovs only (~5% of fovs, ~3% of channels)
Aqua AMSU-A radiances EUMETSAT ATOVS Retransmission Service Radiances for
Tromso, Maspalomas and Edmonton (3 out of 8 stations). RTTOV-7 radiative transfer model replaces RTTOV-5 New bias correction scheme using 850-300hPa and 200-
50hPa thicknesses as predictors rather than AMSU channels 5 and 9.
Use of ATOVS radiances over all land areas (previously restricted to elevations < 1000m)
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
•MODIS imagery from Terra and Aqua used to generate winds.
•IR(11m) and WV (6.7m) channels
•100 min between overlapping images.
•Time delay of 5-6 hours after valid time before winds are available.
•Still experimental. Met Office obtains them via ECMWF. Now using winds from Washington.Picture courtesy of CIMSS
MODIS polar winds
Data Exchange Meeting, Dorval 26-28 May 04
MODIS winds impact
New results using NESDIS-NCEP MODIS windsBetter results than earlier winds from CIMSS
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
II. SSMI TPW used to validate model performance
O - B 22GHz
model‘dry’
model ‘wet’
‘normal’ regime
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
(A)ATSR SSTanomalies
Climate Model SST anomalies
AATSR
ATSR-2ATSRERS-2 gyrofailure
Data Exchange Meeting, Dorval 26-28 May 04
Met Office Report
Relocation Model developments Use of ATOVS AIRS data MODIS atmospheric motion winds SSM/I water vapour Sea surface temperature GPS
Roger Saunders
Presentation to N.America-Europe Data Exchange Meeting,Dorval 26-28 May 2004
Data Exchange Meeting, Dorval 26-28 May 04
GPS Stations for TCWV
Current GPS network shown over Europe
TCWV data put on GTS in real time
Plans to assimilate in Euro model
Data Exchange Meeting, Dorval 26-28 May 04
Ozonesondes from Lerwick The Met Office started launching
weekly ozonesonde ascents from Lerwick in Feb 04
Normally ~11Z on a Wednesday Ascents put on GTS in real time in
WMO BUFR format Data also sent to NILU To ensure continuity an email/letter
of interest to Met Office would help
Data Exchange Meeting, Dorval 26-28 May 04
Future PlansShort term Assimilation of MODIS winds More ATOVS in Mesoscale/EARS in EUROLAM Day-2 AIRS assimilation (Cloud cleared, recon rads…) GPS-RO trials using CHAMP (initial results
encouraging)
Longer term: Assimilation of more data over land Assimilation of cloudy radiances Assimilation of precip inferred from microwave
radiances Use of METOP data
Data Exchange Meeting, Dorval 26-28 May 04
Advanced sounder data volumes
1
10
100
1000
10000
100000
Total received (2003) ATOVS 3X received AIRS received IASI compressed IASI uncompressed
Mb
yte
s/d
ay
Current plans