© Crown copyright 2004 Page 1 11.30 local satellite image Use of microwave radiances for weather forecasting Roger Saunders SFCG-24 20 Sep 2004
© Crown copyright 2004 Page 1
11.30 local satellite image
Use of microwave radiances for weather forecasting
Roger Saunders
SFCG-24 20 Sep 2004
© Crown copyright 2004 Page 3
Cloud is commonBand Instrument Cloud-free Cloud-free upper-trop
INFRARED AIRS (14 km) 5% 30%
MICROWAVE AMSU (50 km) 70% 95%
© Crown copyright 2004 Page 4
In situ vs satellite data coverage
Radiosondes Pilots and profilers Aircraft
Synops and ships Buoys
ATOVS Satobs Geo radiances
ScatterometerSSM/I Ozone
Data coverage09 – 15 UTC5 September 2003 +AQUA (Airs,AMSUA) and 5 geo rads.
© Crown copyright 2004 Page 5
The Advanced Microwave Sounding Unit
Microwave sounders have become very important to the accuracy of NWP
High information content in persistently cloudy data sparse areas.
AMSU is the current operational generation
© Crown copyright 2004 Page 6
© Crown copyright 2004 Page 7
Global compositeof brightness
temperature (K) from
AMSU-AChannel 3
© Crown copyright 2004 Page 8
AMSU cloud liquid water index
The AMSU window channelsindicate the location of thehighest liquid water contentsThese occur near the stormhead and also along the trailing cold frontThe infra-red imagery givesan indication of this to theexperienced eye but is not soquantitatively precise
© Crown copyright 2004 Page 9
Microwave spectrum used by Met Satellites
© Crown copyright 2004 Page 10
Sensitivity of brightness temperature to geophysical parameters over land surfaces
5 10 15 20 25 30 35 40Frequency (G H z)
Soil m o isture
Vege tation biom assS urface roughness
Integrated w ater vapour
C loud liquid w ater
TbPi
© Crown copyright 2004 Page 11
Current use of microwave spectrum for meteorology
AMSU-A/BAtmospheric temp, water vapour + surface
300089 + 150 + 157AMSU-AAtmospheric temp 500050-55
AMSU-BAtmospheric temp, water vapour + surface
14000183+/-7
AMSU-AAtmospheric temp, liquid water
50031.4
SSM/I, AMSU-A
Water vapour, liquid water, atmos sounding
40023.8
AMSR, TMI, SSM/I
Rain rate, ice, water vapour, wind speed
20018.7
AMSR, TMIRain rate, snow, ice, SST, ocean wind
10010.65AMSRSST2006.92
SensorMeasurementBandwidth (MHz)
Frequency (GHz)
© Crown copyright 2004 Page 12
Weighting functions
Emission
Emission
Emission
Atte
nuat
ion Space
Surface
WF(p).T(p) dp∫
Observed brightness temperature (TB)
=
Weighting function (WF)
Pre
ssur
e
© Crown copyright 2004 Page 13
Use of 24 GHz channel for weather forecasting
NWP ‘signal’ at 24GHz in degK
Per
cent
age
of d
ata
belo
w R
FI th
resh
old
of 0
.2K
• RFI would exceed 0.2K threshold 38% of the time over land.• A threshold of 0.03K is required to reduce significant interference to < 5% over land
24 GHz channel is used for identifying best footprints for sounding channelsAlso used for inferring water vapour, cloud water and surface properties
© Crown copyright 2004 Page 14
Microwave imagers: SSM/I, AMSR, TMI…
Wind speed information from vertical and horizontal polarised microwave radiometer data is well established.
© Crown copyright 2004 Page 15
Wind Scatterometers
• Calibrated radar operating at frequencies which are only weakly affected by atmosphere
• Measures surface backscatter at several azimuth angles
Sea-ice, Snow
Sea-surface winds
© Crown copyright 2004 Page 16
Precipitation from microwave radiometers
Scattering by rain at85 GHz showsrain bands clearly.
© Crown copyright 2004 Page 17
Forecast improvements
ECMWF forecasts 1981-2003Fo
reca
st s
kill
Year
© Crown copyright 2004 Page 18
Forecasts from all NWP centres
Annual-mean r.m.s. errors against analyses from WMO scores500hPa height (m) Northern hemisphere
D+3
D+4
D+5
Recent improvement in the accuracy of forecasts
© Crown copyright 2004 Page 19
Forecast skill vs time
© Crown copyright 2004 Page 20
Satellite vs conventional: NH height
Sat data largest impact ~10 hr gain
Forecast skill
© Crown copyright 2004 Page 21
Conventional vs satellite: SH height
Sat data largest impact ~48 hr gain
Forecast skill
© Crown copyright 2004 Page 22
Comparison of overall impact of observing systems0
1020
30
Analysis Observations
No Satellite
No SatelliteSoundingsNo Satellite winds
No Scatterometer
Losing 2 out of 3ATOVSLosing 1 out of 3ATOVSNo Radiosonde
No radiosonde Tand qNo aircraft
Glo
bal N
WP
sco
re
…surface pressure obs also very important to prevent drift in mass of atmosphere
© Crown copyright 2004 Page 23
Impact
The additional information provided by the observations is very small in any given assimilation cycle - does it actually have any impact?
1-2 days forecast improvement in the SH 1/3-1/2 days forecast improvement in the NHHumidity information is most important in the tropics
AMSU is the most important current source of satellite data, providing most of this impact.
© Crown copyright 2004 Page 24
Thanks and please keep the channels we use
free of interference.
© Crown copyright 2004 Page 25
ATOVS & 3D-varATOVS & 3D-var
Direct radiance assimilation Direct radiance assimilation