rttov performance evaluation - ecmwf · qc bc initial fields x simulation b (h(x)) o‐b increments...
TRANSCRIPT
RTTOV Performance Evaluation
Ma Gang
National Satellite Meteorological Center
30,Apr. 2019
• Fast Radiative Transfer Model
• RTTOV to FY satellite
• Pre-Quality Control to assimilation of FY Data
• Summery
• Program in future
QC
BC
Initial fields X
Simulation B(H(X)) O‐B
IncrementsAdjusted initial
FiledsAnalysis
X Predition
Observation Oor yo
NWP Space
Forward OperatorHMap to Observation Space
Adjoint to Forward Operator HT
Map to Model Space
Observation Space
Fast Radiative Transfer Model
Fast Radiative Transfer Model
P(1)
P(n)
T(p) q(p) O3(p)
Ts, qs, Ps, s
R1, R2, R3... Optical Depth
cloud
R1:Downwelling and reflected by
cloud/surface
R2: Emission from surface
R3: Emission from atmosphere
R2
R1
R1
R3
LBL model
Radiance & transmittances
Profiles to absorbers
Amount of absorbers
Temperature profiles
Pressure profiles
Strength of absorb
line Shape of absorb
line
HITRAN,GEISA
LBL
660 665 670 675 6801E-26
1E-25
1E-24
1E-23
1E-22
1E-21
1E-20
1E-19
1E-18
Line
inte
nsity
wave number(cm-1)
660-680cm-1:3025 absorb lines
NeδT of HIRAS
2287channels
2200 2250 2300 2350 2400 2450 2500 25500
0.5
1
1.5
2
2.5
3
3.5
4
Wavenumber/cm-1
NE
dT/K
Forward NEdT of pipe 9 at Tem of 280K
650 700 750 800 850 900 950 1000 1050 11000
0.2
0.4
0.6
0.8
1
1.2
Wavenumber/cm-1
NE
dT/K
Forward NEdT of pipe 1 at Tem of 280K
1250 1300 1350 1400 1450 1500 1550 1600 1650 1700 17500
0.5
1
1.5
2
2.5
Wavenumber/cm-1
NEdT
/K
Forward NEdT of pipe 5 at Tem of 280K
What is RTTOV
• RTTOV : Radiative Transfer model for TOVS, UK MetOffice, 1993
• TOVS: TIROS Vertical Sounder
sunsunss
pd
s
p
uss
FpTTdpppT
pBT
dpppT
pBTBR
s
s
cos),(),(
))(()1(
),())(()(
,0,0
*
,
0
,
d: optical depth to satellite channels
X: predictors to d and depend on atmosphere status
a: coefficients to d and response to spectral characters
of channels
Predictors – RTTOV v7
• RTTOV to FY satellite
RTTOV to FY satellite
Started at Lannion, 1999 • GENLN2, TIGR43,NESDIS34
• RTTOV5
• VIRR of FY1c,VISSR of FY2b
RTTOV for infrared sensors of FY3
• IRAS 20 infrared channels• Transmittances data base in 0.5
cm-1 resolution• Transmittances data base from
600cm-1 – 3000cm-1• TIGR43 profiles to generate
coefficients• NESDIS34 to do independent test 0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
channel
STD Brightness Temperature (K
RTTOV for microwave sensors
• MPM LIEBE89/93• MONORTM
174
176
178
180
182
184
186
188
190
192
194
0 10 20 30 40 50
频点数
频率(GHx) 183.31+-1GHz
183.31+-3GHz
183.31+-7GHz
Wavenumber
Filter response
Radiance
• Planck weighted convolution to 4.3μm channels
mean STD
dBF
ddBFB
dBF
ddBFB
dBF
dBFB
dF
dLF
T
TTs
T
TT
T
sTsTclr
z
z
,
1
,,,
,
1
,,
,
,,,,
,
,
)1(
To improve accuracy of fast model
Before correction
After correction
1: VISSR/FY2E2: VISSR/ FY2F3: SERVIRI /MSG14: IRAS/FY3B
Analysis to predictors of water line absorption
RTTOV v7Predictors:13
RTTOV v8Predictors:12
RTTOV v9Predictors:19
CRTM 2.1Predictors:14
)()( jWSec r)()( jWSec r
)())()((
*
2
jWjWSec r
)()()( jTjWSec r
)()( 22 jWSec r
)())()(( jTjWSec r 4 )()( jWSec r
)()())()((
* jWjWjWSec rr
3))()(( jWSec r4))()(( jWSec r
)()()()( r jTjTjWSec 4
* ))()(( jWSec 2
* ))()(( jWSec
)()( 22 jWSec r)()( jWSec
2)()( jWSec
)()()( jTjWSec r
)()( jWSec r
4 )()( jWSec r
)()( jWSec r3))()(( jWSec r
)()()()( r jTjTjWSec
)())()(( jTjWSec r
(j)W
)()(2
jWSec r
(j)W
)()(3
jWSec r
)()( jWSec r)()()( jTjWSec r
)()( 22 jWSec r)()()()( r jTjTjWSec
4 )()( jWSec r3))()(( jWSec r
)()( jWSec r)())()(( jTjWSec r
4))()(( jWSec r
(j)W
)()(
jWSec r
(j)W
)()(3
jWSec r
22 )()( jWSec )()( jWSec
)(Sec
)()( 22 jWSec r)()( jWSec
22 )()( jWSec )()()( jTjWSec r
)()( jWSec r4 )()( jWSec r
)()( jWSec r3))()(( jWSec r
4))()(( jWSec r)()()()( r jTjTjWSec )())()(( jTjWSec r
(j)W
)()(2
jWSec r
(j)W
)()(3
jWSec r
)()(3jWSec
)()(3jWSec r4 )()()( jWjWSec
)()()( jWjWSec rr
)()()()()( jTjWSecjWSec rr
(j)W
)()(3
jWSec r
Coefficients of AGRI to various optical depth predictor
LBL:GENLN2(Mixed gas,Water-line,Ozone),CKD2.1(Water Continum)Profile Data: NESDIS 34Layers: 43,1013.25-0.1hPa
Profiles database
Distributoin of TIGR43
#*#*#*#*
#*#*
#*#*
#*#*#*
#*
#*
#*#*#*
#*
#*
#*#*
#*
#*
#*
#*#* #*
#*
#*
#*#*
30 typical profiles in China
ID of layer
Type 1 (hPa)
Type 2 (hPa)
Type 3 (hPa)
Type 4 (hPa)
Type 5 (hPa
1 20 20 20 20 20
2 30 30 30 30 30
3 50 50 50 50 50
4 70 70 70 70 70
5 100 100 100 100 100
6 150 150 150 150 150
7 200 200 200 200 200
8 250 250 250 250 250
9 300 300 300 300 300
10 400 400 400 400 400
11 500 500 500 500 500
12 700 700 700 700
13 850 850 850
14 925 925
15 1000
TibetanLoess plateau
Traditional plain
190 200 210 220 230 240 250 260 270 280 290
1
2
3
Temperature Profile 1-5
Temperature(K)
Pre
ssur
e(hP
a)
pro1pro2pro3pro4pro5
180 200 220 240 260 280 300
101
102
103
Temperature Profile 6-10
Temperature(K)
Pre
ssur
e(hP
a)
pro6pro7pro8pro9pro10
200 220 240 260 280 300 320
101
102
103
Temperature Profile 11-15
Temperature(K)
Pre
ssur
e(hP
a)
pro11pro12pro13pro14pro15
200 210 220 230 240 250 260 270 280 290 300
101
102
103
Temperature Profile 16-20
Temperature(K)
Pre
ssur
e(hP
a)
pro16pro17pro18pro19pro20
180 200 220 240 260 280 300
101
102
103
Temperature Profile 21-25
Temperature(K)
Pre
ssur
e(hP
a)
pro21pro22pro23pro24pro25
180 200 220 240 260 280 300
101
102
103
Temperature Profile 26-30
Temperature(K)
Pre
ssur
e(hP
a)
pro26pro27pro28pro29pro30
Temperature gap to typical profiles in China Gap to troposphere top Gap to temperature at same pressure level Gap to temperature inversion layer
Humidity gap to typical profiles in China
LBLRTM整层大气透过率
0.00E+00
1.00E-01
2.00E-01
3.00E-01
4.00E-01
5.00E-01
6.00E-01
7.00E-01
8.00E-01
9.00E-01
1.00E+00
2180 2185 2190 2195 2200 2205 2210 2215 2220 2225 2230
wavenumber(cm-1)
tran
smi
ttan
ce
Transmittances data base (GENLN2 vs LBLRTM)GENLN2整层大气透过率
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
800 805 810 815 820 825 830 835 840 845 850
wavenumber(cm-1)
transmittance
0度 36.87度
48.17度 55.15度
60度 63.61度
GENLN2整层大气透过率
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
2180 2185 2190 2195 2200 2205 2210 2215 2220 2225 2230
wavenumber(cm-1)
transmittance
0度 36.87度
48.17度 55.15度
60度 63.61度
LBLRTM整层大气透过率
0.00E+00
1.00E-01
2.00E-01
3.00E-01
4.00E-01
5.00E-01
6.00E-01
7.00E-01
8.00E-01
9.00E-01
1.00E+00
800 805 810 815 820 825 830 835 840 845 850wavenumber(cm-1)
transmittance
Total transmittances
Total transmittances
Total transmittances
• Pre-Quality Control to assimilation of FY Data
Cloud is commonInstrument Cloud-free Cloud-free upper-trop
AIRS (14 km) 5% 30%
AMSU (50 km) 70% 95%
Window 89.015
O2f0±0.3222±0.04514
O2f0±0.3222±0.01013
O2f0±0.3222±0.02212
O2f0±0.3222±0.04811
O2f0±0.21710
O2f0=57.29±0.3449
O255.58
O254.947
O254.46
O253.596+/0.1155
O252.84
Window 50.33
Window 31.42
H2O23.81
absorberCnetral frequency(GHz)
ID
AMSU-A
IDCentral frequency
(GHz)
3dBband width
(MHz)
Use
1 50.3 180Emissiv
ity
2 51.76 400 Sounding to
atmospheric
temperature
3 52.8 400
4 53.596 400
5 54.40 400
6 54.94 400
7 55.50 330
8 57.290344(fo) 330
9 fo±0.217 78
10 fo±0.3222±0.048 36
11 fo±0.3222±0.022 16
12 fo±0.3222±0.010 8
13 fo±0.3222±0.0045 3
MWTS Ⅱ
微波波段温度探测仪器
ID Central frequency(GHz) absorbers
1 50.3 Window
2 53.596±0.115 O2
3 54.94 O2
4 57.290 O2
MWTS
Pre-Quality Control to assimilation of FY Data
VIRR
MWTSⅡ
t mX LX LX
t mY LY LY
1) X: deviation to centers between matched foot-point and target foot-pointX > a•LX
2) Y: deviation to scanline between matched foot-point and target foot-pointY > b•LY
3) Sum to deviation from center of matched foot-point to two focus of target foot-point
1 2 2 2C O C O
Pre-Quality Control to assimilation of FY Data
cloud clear
Cloud mask by VIRR
Cloud fraction after matched onto foot-point of MWTS
𝐿 1 𝑁 · 𝐿 𝑁 · 𝐿
N: cloud fractionL: radiance in a foot-point
A: QC with O-B onlyB: QC with both O-B and cloud fraction
Sensitivity between radiance and cloud fraction
O-B:O-B:
O-B to MWTSII of FY3C
P(1)
P(n)
T(p) q(p) O3(p)
Ts, qs, Ps, s
R1, R2, R3... Optical depth
Scattering & emission
Pre-Quality Control to assimilation of FY Data
𝐿 𝐿 𝑑𝑝 𝐿 𝑑𝑝 𝐿 𝑑𝑝
H
Channel selection
Multi-layers fast forward model in RTTOV
Efficient cloud fraction at each layer
cldtop
s
cldtop
s
P
P
iclditop
cldi
P
P iicld
topcld
itopcldi
cldi dp
PTB
dpP
TBTBL
22
,,)(
)()1()()(
Radiance to cloudy/partly cloudy pixel
Radiance to multi-layer cloud
k
j
cldij
clrii LNLNL
1max )1(
Reflection at cloud top
Maximum cloud fractions at each layer
Clear radiance above cloud top
Pre-Quality Control to assimilation of FY Data
53.596GHz 54.4GHz
54.94GHz 55.5GHz
Pre-Quality Control to assimilation of FY Data
CTRL
0.10K
0.05K
0.20K
Pre-Quality Control to assimilation of FY Data
O-B to 3 channels
Data used in 3 experiments
Summery
1, RTTOV has been used in application of FY data at CMA with various
version
2, Coefficients to RTTOV could be generated to sensors of FY satellite both in
infrared and in microwave at NSMC
3, Satisfied simulations to FY satellite by RTTOV could be obtained while
comparisons are performed to radiance not only from LBL model, but also
from observations
4, Pre-quality controls to MWTS have been used that is based on analysis to
sensitivity between radiance and cloud parameters by RTTOV
Future : AI in radiative transfer
Deviation between AI and RTTOV
• Profiles dataset to training coefficients were classified by total water vapor content
STD to wet profiles STD to dry profile
Future : non-unfied training database
Future : ray-tracing in radiative transfer calculation
P(1)
P(n)
Ts, qs, Ps, s
R
𝑅 𝜀 · 𝐵 𝑇 · 𝜏 𝐵 𝑇 , ·𝜕𝜏 ,𝜕𝑝 𝑑𝑝 1 𝜀 · 𝜏 𝐵 𝑇 , ·
𝜕𝜏 ,𝜕𝑝 𝑑𝑝
n-grids: number of grids from center of foot-point at
surface to project of the center at a pressure layer
Thanks for your attention