Nudged isotope AGCM simulation and its comparison with TES, SCIAMACHY and

ground-based FTS isotope retrievals

Kei YoshimuraScripps Institution of Oceanography

Collaborators: TES (J. Worden, D. Noone)

SCIAMACHY (C. Frankenberg, T. Rockman)Ground based FTS (M. Schneider, F. Hase, J. Notholt)

TES workshop in Boulder, Feb. 23-25, 2009

Contents

• Model – Ground based FTS comparison

• Model – Satellite comparison (UT and PBL)

• Isotope Assimilation (plan)

Issues in GCM vs RS comparison

• Timing (clear sky vs rainy)• Space (vertical and horizontal)

• Significant difficulty: GCM’s time and “real” time is different!

A solution: Spectral Nudging– Poor man’s data assimilation –

IsoGSMSIO/UT Isotopic Global Spectral Model

H18OH

HD16O

( )⎩⎨⎧

≥+×−

Comparison with GNIP Climatology“Rainy sky integration”

Ny Alesund

Kiruna

Bremen

Izana@2370m

-350

-300

-250

-200

-150

-100Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

IsoGSMclim Bremen_FTSIsoGSMclim Kiruna_FTSIsoGSMclim NyAlesund_FTSIsoGSMclim Izana_FTS

Monthly climatology of vapor dD

Comparison with FTIR Climatology“Clear sky integration”

Collocated: Ny Alesund

-500

-450

-400

-350

-300

-250

-200

-150

-10001/11/5 02/7/13 03/3/20 03/11/25 04/8/1 05/4/8 05/12/14 06/8/21

FTSIsoGSM_TPW R=0.68

-450

-400

-350

-300

-250

-200

-150

-10095/10/28 98/7/24 01/4/19 04/1/14 06/10/10

FTSIsoGSM_TPW

Kiruna-350-300

-250

-200

-150

-10002/1/1 02/4/11 02/7/20 02/10/28 03/2/5 03/5/16 03/8/24 03/12/2

FTSIsoGSM_TPW

R=0.77

Bremen FTS

y = 0.5304x - 75.269R2 = 0.4988

-350

-300

-250

-200

-150

-100

-50-350 -250 -150 -50

FTS

IsoG

SM

-350

-300

-250

-200

-150

-100

-50

004/6/12 04/12/29 05/7/17 06/2/2 06/8/21 07/3/9 07/9/25

FTSIsoGSM_TPW

R=0.71

-450

-400

-350

-300

-250

-200

-150

-100

-5098/11/1 00/3/15 01/7/28 02/12/10 04/4/23 05/9/5 07/1/18 08/6/1

FTSIsoGSM_700hPa

Izana-450-400

-350

-300

-250

-200

-150

-100

-5004/1/1 04/4/10 04/7/19 04/10/27 05/2/4 05/5/15 05/8/23 05/12/1

FTSIsoGSM_700hPa

R=0.61

-350

-300

-250

-200

-150

-100Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

IsoGSMclim Bremen_FTSIsoGSMclim Kiruna_FTSIsoGSMclim NyAlesund_FTSIsoGSMclim Izana_FTS

-350

-300

-250

-200

-150

-100Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

IsoGSMclim Bremen_FTS IsoGSMIsoGSMclim Kiruna_FTS IsoGSMIsoGSMclim NyAlesund_FTS IsoGSMIsoGSMclim Izana_FTS IsoGSM

Collocated model vs FTS

y = 1.4331x + 2.5103

R2 = 0.69

0

5

10

15

20

25

0 5 10 15 20 25

y = 0.8714x - 19.949

R2 = 0.301

-350

-300

-250

-200

-150

-100

-50

-350 -300 -250 -200 -150 -100 -50

y = 0.9455x + 15.704

R2 = 0.9415

250

255

260

265

270

275

280

285

250 255 260 265 270 275 280 285

0

5

10

15

20

25

1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235 244 253 262 271 280 289 298 307 316 325 334 343 352 361 370 379 388 397 406 415

-350

-300

-250

-200

-150

-100

-50

1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235 244 253 262 271 280 289 298 307 316 325 334 343 352 361 370 379 388 397 406 415

250

255

260

265

270

275

280

285

1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235 244 253 262 271 280 289 298 307 316 325 334 343 352 361 370 379 388 397 406 415

Temperature

Water Vapor

Vapor HDO

2004/09/20-21

Pink: Simulation, Violet: Observation

Obs

Sim

TES vs collocated-IsoGSM comparison

Correlation b/w TES & IsoGSM(averages of 2004-2007 monthly score)

Accuracy order:• T > Humidity > HDO• For HDO

– N-hemi > S-hemi– Mid > Low > High

00.10.20.30.40.50.60.70.80.9

1

N-hi N-mi N-lo S-hi S-mi S-lo0

0.10.20.30.40.50.60.70.80.9

1

N-hi N-mi N-lo S-hi S-mi S-lo

00.10.20.30.40.50.60.70.80.9

1

N-hi N-mi N-lo S-hi S-mi S-lo

Temperature Humidity

HDO in Water Vapor

Yoshimura et al., in prep

SCIAMACHY HDO retrieval (Frankenberg et al., 2009)It sees the boundary layer!

-190

-170

-150

-130

-110

-90

-70

-50Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

IsoGSMclim Izana_SCIA IsoGSMcol

SCIA/FTS/collocated-IsoGSM comparison at Izana(left: SCIA vs Model, right: FTS vs Model)

-240

-220

-200

-180

-160

-140

-120

-100Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

IsoGSMclim Izana_FTS IsoGSMcol

FTS@2370m, IsoGSM@700hPa

SCIA@clm vs IsoGSM@clm

Yoshimura et al., in prep

Difference in “measurements” for DJF-JJA climatology

GN

IP (r

ain)

SC

IAM

AC

HY

(col

umn

vapo

r)TE

S(5

00-8

25hP

a)

Frankenberg et al., submitted

dD in H2O

Rai

nco

lum

n va

por

500-

825h

Pa

Model (IsoGSM) resultsDifference in “measurements” for DJF-JJA climatologyG

NIP

(rai

n)SC

IAM

ACH

Y(c

olum

n va

por)

TES

(500

-825

hPa)

Science: Assimilation of isotopes– No longer a poor man! –

Observation

Model

Assimilation

IsoGSM(Yoshimura et al., 2008)

What is appropriate method?

TES/Aura HDO(Worden et al., 2007)

SCHIAMACHY(Frankenberg et al., 2009)

FTS

• 1st Purpose: Make dD vapor analysis fields

How about EnKF?

• In comparison with variational methods:– Better than 3D-Var in general. – Adjoint model unnecessary. (needed by 4D-

Var)• Possibility to improve (influence) other

thermo-dynamical fields (q, wind, T, etc.) 2nd Purpose (for better weather analysis/forecast, maybe)

(c.f., wind field improvement by humidity assimilation; Liu et al., 2008)

General form of Kalman Filter

( )( )

( ) fiiiaii

Ti

fii

Ti

fii

fii

oii

fi

ai

Tai

fi

ai

fi

H

M

PHKIP

RHPHHPK

xyKxx

MMPP

xx

−=

+=

−+=

=

=

−

−

+

1

1

1

)(

)(x: state in model’s worldy: state by observationP: model error covarianceR: obs error covarianceM, M: modelH, H: obs operatorK: Kalman gain

Forecast model

Forecast of error covariance

Kalman filter

Kalman gain

Analysis of error covariance

Ensemble Kalman Filter(Square Root Filter; SRF)

[ ]

fT

fT

T

foT

fT

ff

afofa

Tf

Tf

Tf

m

mm

H

H

XHδRXHδIUDUUUD

XyRXHδUUDδXX

δXXyKXX

RXHδUUDδXK

δXδXPXXδX

xxX

1

2/1

11

11

1-

1

)()1( where]1

))(()([

))((

)(

1)-(m

−

−

−−

−−

+−=

−+

−+=

+−+=

=

=

−=

= L Ensemble Matrix (small x means each ensemble member) Ensemble perturbation (anomaly)

Eigenvalue decomposit

Kalman gain

Ensemble analysis

Perturbation Model error covariance

x: state in model’s worldy: state by observationP: model error covarianceR: obs error covarianceM, M: modelH, H: obs operatorK: Kalman gain

SSI

Obs.Data

GSMSSI

Obs.Data

GSMSSI

Obs.Data

Analyzed fields

Forecasted fields Forecasted fields Forecasted fields

Analyzed fields Analyzed fields

Forecasted fields ofEnsemble members

EnKF

Obs.Data

Analyzed fields ofEnsemble members

Model

Forecasted fields ofEnsemble members

EnKF

Obs.Data

Analyzed fields ofEnsemble members

Model

Forecasted fields ofEnsemble members

EnKF

Obs.Data

Analyzed fields ofEnsemble members

Framework of “Isotope Assimilation”

TES HDO obs.data

H18OH

HD16O

IsoGSM

LETKF

[ ]

afofa

Tf

Tf

m

H δXXyKXX

RXHδUUDδXK

XXδXxxX

+−+=

=

−==−−

))((

)(

,11

1 L

Ensemble Forecasting

3rd Purpose: Estimate “required” accuracy and resolution of HDO measurement for “adequate” isotope assimilation

Nudged isotope AGCM simulation and �its comparison with TES, SCIAMACHY and ground-based FTS isotope retrievalsIssues in GCM vs RS comparisonA solution: Spectral Nudging�– Poor man’s data assimilation –Comparison with GNIP Climatology�“Rainy sky integration”Comparison with FTIR Climatology�“Clear sky integration”Collocated: Ny AlesundBremen FTSCollocated model vs FTSTES vs collocated-IsoGSM comparisonCorrelation b/w TES & IsoGSM�(averages of 2004-2007 monthly score)SCIAMACHY HDO retrieval (Frankenberg et al., 2009)�It sees the boundary layer!SCIA/FTS/collocated-IsoGSM comparison at Izana�(left: SCIA vs Model, right: FTS vs Model)Science: Assimilation of isotopes�– No longer a poor man! –How about EnKF?General form of Kalman FilterEnsemble Kalman Filter�(Square Root Filter; SRF)Framework of “Isotope Assimilation”

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure true /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles true /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /NA /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /LeaveUntagged /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice