RTTOV Performance Evaluation - ECMWF · QC BC Initial fields X Simulation B (H(X)) O‐B Increments Adjusted initial Fileds Analysis X Predition ObservationO oryo NWP Space Forward

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


1250 1300 1350 1400 1450 1500 1550 1600 1650 1700 17500

0.5

1

1.5

2

2.5

Wavenumber/cm-1

NEdT

/K


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



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(K)

Pre

ssur

e(hP

a)


200 220 240 260 280 300 320

101

102

103


Temperature(K)

Pre

ssur

e(hP

a)


200 210 220 230 240 250 260 270 280 290 300

101

102

103


Temperature(K)

Pre

ssur

e(hP

a)


180 200 220 240 260 280 300

101

102

103


Temperature(K)

Pre

ssur

e(hP

a)


180 200 220 240 260 280 300

101

102

103


Temperature(K)

Pre

ssur

e(hP

a)


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



• 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


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


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


53.596GHz 54.4GHz

54.94GHz 55.5GHz


CTRL

0.10K

0.05K

0.20K


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

Documents

RTTOV Performance Evaluation - ECMWF · QC BC Initial fields X Simulation B (H(X)) O‐B Increments Adjusted initial Fileds Analysis X Predition ObservationO oryo NWP Space Forward