Operational Use of the ATOVS radiances in Masahiro Kazumori

advertisement
Operational Use of the ATOVS radiances in
global data assimilation at the JMA
Masahiro Kazumori
Kozo Okamoto,Hiromi Owada
Japan Meteorological Agency(JMA)
Numerical Prediction Division / Japan Meteorological Agency
Recent Progress of Satellite Data
Assimilation in JMA
Global data assimilation
l
l
l
Sep 25 2001 implementation of 3D-Var global data assimilation
May 6 2003 use of QuikScat sea surface wind data
May 6 2003 use of Meteosat high-density atmospheric motion
Wind data
l May 28 2003 start of operational direct assimilation of NOAA15&16
ATOVS radiance + new cumulus parameterization scheme
Previous Status
Retrievals use
NESDIS retrieved
thickness
GMS-5 moisture data
ls
a
v
ie
r
t
re
l
a
ov
m
Re
Current Status
Radiances use
NESDIS Level 1D TBB
(HIRS/3, AMSU-A)
NESDIS Level 1C TBB
(AMSU-B)
Less Conversion Error
Numerical Prediction Division / Japan Meteorological Agency
Processing of ATOVS radiances
NESDIS
LevelLevel-1D
HIRS/3,AMSUA
Thinned levellevel-1C
AMSUAMSU-B
GTS&Internet
JMA
1D-Var as Pre-processor
·
·
·
·
·
Thinning
Quality Check
Bias Correction
Channel Selection
Obs Error Assignment
levellevel-1B
AMSUAMSU-A,AMSUA,AMSU-B
Not in use
J= (x(x-xb)
(x-xb)TB-1(x+ (y(y-H(x))
(y-H(x))TR-1(y•
•
•
Global Analysis with 3D-Var
X:analysis variables
– T & lnQ at 41 levels of GSM and Tskin
Y : observation
– HIRS/AMSUA/AMSUB TBB from
NESDIS BUFR
– AMSUA is mapped to HIRS
H : observation operator
– RTTOVRTTOV-6 (Saunders et al.,2000)
· RTTOV-6
· Incremental 3D-Var(Inner T106L40)
· Model Top 0.4hPa
Global Spectral Model(T213L40)
Numerical Prediction Division / Japan Meteorological Agency
Change of distribution of available data
Thinning
•
Constant distance
–
–
•
Temperature(thickness) retrieved by NESDIS
250km(HIRS/AMSA)
180km(AMSUB)
Priority is given to
–
–
clear radiances
satellite closer to
analysis time
Use of data on Land in the upper air.
Numerical Prediction Division / Japan Meteorological Agency
JMA TBB Bias Correction scheme
BIASj (n) = a j 0 + ∑i =1 a ji (n) X ji (n)
5
BIAS=<TBobs- TBcal>
global mean and 1-Year averaged dataset
TBobs: Observed TBB(Collocated with RAOB within 2degree & 90min)
TBcal: Calculated from profiles of RAOB
RAOB are used due to the presence of a model bias
Guess is used to complement the lack of the upper stratosphere temp.
temp. & moist.
Predictors X
AMSU-A Ch5
Calculated TBB
AMSU-A Ch7 Calculated TBB
AMSU-A Ch10 Calculated TBB
TPW(Total Precipitable Water from first guess)
Surface Temperature(JMA SST Analysis)
The coefficients was defined every scan position.
(There are 56scan position due to level 1D data mapped to HIRS Spot)
Not bias-correct AMSU-A12-14, AMSU-B, HIRS11-12 due to large model bias
Numerical Prediction Division / Japan Meteorological Agency
Complicated Scan Bias of Level 1D data
With BIAS Correction
W/O BIAS Correction
Effect of
Mapping to
HIRS Spot
AMSU-A Ch4
AMSU-A Ch5
AMSU-A Ch6
AMSU-A Ch7
AMSU-A Ch8
AMSU-A Ch9
AMSU-A Ch10
AMSU-A Ch11
From:2003/06/27
To:2003/06/30
AMSU-A Ch12
AMSU-A Ch13
AMSU-A Ch14
Numerical Prediction Division / Japan Meteorological Agency
Cycle Experiments
Period
2001 December
2002 July
Model
JMA Global Spectrum Model : T213L40
Assimilation system : 3D-Var with RTTOV-6
Setting
CNTL ATOVS Retrievals assimilation
TEST ATOVS radiance assimilation
NOAA15,NOAA16, HIRS/3,AMSU-A,AMSU-B
Numerical Prediction Division / Japan Meteorological Agency
Change of Analysis Increment
Temperature
(Zonal mean)
CNTL
TEST
d !!
a
e lly
r
Sp ica
t
r
ve
Moisture 850hPa
(Specific Humidity)
CNTL
TEST
!!
y
ll
a
b
o
Gl
Numerical Prediction Division / Japan Meteorological Agency
Analysis Impact for Temperature
Zonal Averaged Monthly Mean July 2002
CNTL
Verification against RAOB
Temperature BIAS Error
20S-20N
TEST
CNTL
TEST
Smooth!!
Decrease of Bias!!
Numerical Prediction Division / Japan Meteorological Agency
Analysis Impact for Humidity
Effect of ATOVS moisture channels
(HIRS/3:ch10,ch11,ch12 & AMSU-B:ch3,ch4,ch5)
Validation against Large impact on Total precipitable water
in tropics was found.
SSM/I TPW
Difference(CNTL-SSM/I)
N.H.
Red:TEST - SSM/I
Blue:CNTL - SSM/I
Trop
S.H.
Difference(TEST-SSM/I)
(2002/07/15 12UTC)
Humidity field is close to
SSM/I observation and
became realistic!!
Numerical Prediction Division / Japan Meteorological Agency
Forecast Impact
500hPa Geopotential Height
Verification against RAOB
CNTL
Jul2002
Jul2002
RMSE
BIAS
RMSE
BIAS
TEST
Dec2001
Dec2001
RMSE
BIAS
RMSE
BIAS
N.H.
Positive
Tropics
Impact!!
S.H.
Numerical Prediction Division / Japan Meteorological Agency
RMSE of 500hPa geopotential height
improved especially in former part of
Forecast time.
RMSE Difference(TEST-CNTL)
24hour Forecast(30days ave.)
Dec2001
Dec2001
TEST
CNTL
TEST
CNTL
RMSE(m)
l
a
ob
l
G
Jul2002
Jul2002
RMSE(m)
Forecast Score Z500
(against Initial)
l
a
ob
l
G
TEST
CNTL
TEST
CNTL
Impact is remarkable on
Southern Hemisphere in
24hour forecast!!
Numerical Prediction Division / Japan Meteorological Agency
Case Study: Typhoon Track Prediction
500hPa geopotential height
Analysis
27 May 2003 12UTC INIT
Forecast(TEST)
Difference
Forecast(CNTL)
Difference
Analysis
TEST
CNTL
Tip of arrow means
36hour forecast
Forecast of strength of subtropic high pressure
affects forecast of typhoon track prediction.
Numerical Prediction Division / Japan Meteorological Agency
Comparison with Other NWP Center
JMA: Direct assimilation ATOVS radiances
JMA: Retrievals assimilation
ECMWF
UKMO
NCEP
CANADA
Improvement
of 24hour
forecast is
remarkable!!
RMSE of Global 500hPa Height
Dec 2001
Jul 2002
FT=120
FT=72
FT=24
Numerical Prediction Division / Japan Meteorological Agency
Summary and Ongoing Work
• Since 28 May 2003, JMA has started direct assimilation of ATOVS radiances in
global 3d-Var operationally.
• In the experiments, analyzed temperature and humidity field were verified and
became realistic.
• For forecast skills, very positive impact was found, especially, in tropics and
southern hemisphere at short term forecast.
But, the improvement in northern hemisphere was not large.
Moreover, to improve forecast skills,
• we have some plans and are going to.
– Use level 1B data,
– Revise TBB bias correction scheme,
– Use NOAA 17 (HIRS,AMSU-A,AMSU-B)
– Upgrade radiative tranfer model (RTTOV-6 ---> RTTOV-7)
– Use other new satellite data SSM/I,TRMM,AMSE,AIRS etc.
Numerical Prediction Division / Japan Meteorological Agency
Download