22
Xu Li, John Derber EMC/NCEP Near Sea Surface Temperatures (NSST) Analysis in NCEP GFS

Near Sea Surface Temperatures (NSST) Analysis in … Li, John Derber EMC/NCEP Near Sea Surface Temperatures (NSST) Analysis in NCEP GFS

Embed Size (px)

Citation preview

Xu Li, John DerberEMC/NCEP

Near Sea Surface Temperatures (NSST)Analysis in NCEP GFS

Outline

• What is NSST and why NSST analysis?• How does NSST analysis in NCEP GFS work?• Main characteristics of the new scheme• Progress• Experiments & Results• Plan

(a) Night or Day without diurnal warming (b) Day with strong diurnal warming

Near Sea Surface Temperatures (NSST)

)0( == zTTs

rT

rT

sT

wlT

Warm layer

Sub-layerSub-layer

z

TT

)(zTwl

slT

rz

: cannot be observed directly (SST)

: can be observed directly by satellites: IR & MW (sub-layer T-profile)

: can be observed directly by buoys & ships (warm layer T-profile)

: can be observed directly by buoys & ships (reference T)

cz

cz

)(zTsl

rT

)(zTsl

Diurnal warming profile : (0 ~ 4.0K at z=0),Sub-layer cooling profile : (0 ~ 0.8K at z=0),

)(' zTw

)1(~ mmOzc

)5(~ mOzr

z

NSST Profile : )()()( ''zTzTTzT

cwr!+=

)0()(cslzzTzT !<=

)(rrzzTT ==

)0()(rwlzzTzT !<=

)(' zTc

NSST Analysis in NCEP GFS: Components and Interaction

Radiative Transfer Model

(CRTM)

Air-sea Fluxes

AtmosphericModel

(GFS_AM)

in Sub-layer

SST: )0()0( ''

cw

an

rTTT !+

)()( ''zTzTT

cw

an

r!+=

BG:

AtmosphericAnalysis

(GSI)

NSST Analysis

(GSI)

NSST model

)(' zTw

)(' zTc

IC:

rT

zT

!

! )(

rT

zT

!

! )(

)(zTTr!

csl

cb

T

T

,

,

!

!

cbcslTT,,

!

X

Tcb

!

!,

cbTX,

!

)(zTsl

=sT

)()()( ''zTzTTzT

cw

an

r

an!+=

anX

bgX

NSST Analysis in NCEP GFS: Main Characteristics

• NSST is treated as depthdependent– Advantages

• More realistic thermal boundarycondition to atmospheric modeland radiative transfer model

• Short time scale (hours) air-seainteraction included in weatherforecasting

• Diurnal variation in NSSTresolved

– Work• NSST model and its coupling

with GFS_AM• Observation operator and its

Jacobi for the new analysisvariable and NSST data

• Determination of theobservation depth for NSSTdata (satellite, buoys &ships)

• NSST is analyzed within GSI– Advantages

• 6-hourly analysis• Use more types of data easily

– Better data coverage• Assimilate the raw observations

directly– Simpler observation error

• NSST analyzed withatmospheric analysis variables ina single cost function

– Better consistency– Work

• Incorporation of NSST analysisin GSI

• The use of more data: AVHRR,insitu NSST and more

Progress

• A 6-hourly NSST analysis scheme in NCEP GFS has beendeveloped– Boundary condition (SST) for GFS_AM– Boundary condition (sub-layer T-profile) for CRTM

• A series of experiments on this new scheme have beendone and the results are encouraging– Impact of NSST model on CRTM and therefore atmospheric

analysis (reported last year)– Tr analysis with GSI– Impact of NSST model on GFS_AM (7-day forecasting)– Coupled assimilation (parallel run) in preparation

Experiments of Tr analysis in GSI

IR:

MW:

Used in operation+ AVHRR GAC+ NSST of buoys&ships

OCN

Used in operation+ AVHRR GAC+ NSST of buoys&ships

CTL

Lower thermal boundarycondition to CRTM

Analysisvariable

Observation Data

)()( ''

ircirw

bg

r zTzTT !+

)(' mww

bg

r zTT +

bgSST

rT

SST

)(),( ''zTzT

cw : Provided by uncoupled NSST model (forced byavailable NCEP 3-hourly operational fluxes)

Experiment period: 09/01/2007 ~ 09/07/2007

)0('wT

cwTT !"!

)0('cT

)0()0( ''

cwTT !

:

:

!"

#$%

&

'

'(

'

'))*

+,,-

.(

'

'=

s

Q

s

r

r

h

r

sw

rpo T

I

T

z

z

I

z

I

zcW

/

2

0

!"

#$%

&

'

'+

'

'(()

*++,

-

'

'.=

s

Q

s

r

r

sw

r

h

rpo

dT

I

T

z

z

I

z

I

zcW

22

2/

!"

#$%

&

'

'((+

'

'=

s

cccnscc

s

cT

zzARSQ

T

QzC )(

10 )

*

!!"

#$$%

&

'

'(

'

'=

ss

ccnscd

T

Q

T

zARC )

*

1

Sensitivities of T(z) to Tr

czz !!0z

CW

CW

CWT

Tdd

r

z

000011

1

+!

!+

+!=

"

"

rczzz !!z

CW

W

CW

C

T

Td

r

z

0000

0

11

1

+!+

+!

+=

"

"

Sensitivities of T(z) to Tr

Diurnal variability of analysis variable (Tr)

Diurnal variability of analysis variable (SST)

GDAS analysis (00Z, 06Z, 12Z, 18Z)OCN

GDAS analysis (00Z, 06Z, 12Z, 18Z)CTL

GFS analysis (00Z, 06Z, 12Z, 18Z)GFS

SSTAtmospheric

Initial Conditions

opSST

opSST

)0()0( ''

cw

opTTSST !+

)0()0( ''

cw

opTTSST !+

'',cwTT

: NCEP operational SST analysis

: Provided by coupled model (GFS_AM + NSST model)

Experiments of NSST model on GFS 7-day forecasting.

The experiments (CTL and OCN) have been done for 3 months:July 2007. January 2008 and September 2007.

Plan

• Extend the 7-day forecasting to 14-day• Parallel run

– Feedback between analysis and forecasting• Operational in 2008• Coupling of GFS_AM, NSST model and

OGCM?