Predictability of stratospheric sudden warming events and associated stratosphere-troposphere...

Predictability of stratospheric sudden warming events and associated

stratosphere-troposphere coupling system

T. Hirooka, T. Ichimaru (DEPS, Kyushu Univ.),

H. Mukougawa (DPRI, Kyoto Univ.)

Chapman ConferenceSantorini, 24-28 Sep 2007

Few studies on the predictability in the stratosphere where different dynamics are dominant

A practical predictable period for weather forecasts : 7 days

To examine predictability of Stratospheric Sudden Warming (SSW) events by using the Japan Meteorological Agency (JMA) ensemble one-month forecast model

MotivationObservational errors and imperfection of numerical prediction models

Predictable limit

Contents

• Comparison of the Observational Features of 3 Major SSWs in Dec 2001, Jan 2004 and Jan 2006

• Comparison of the Predictability of 3 Major SSWs

• Statistics of the Predictability of Recent 12 Warmings During 2001/02 and 2005/06 Winters

• Summary

• 1-month forecast data of JMA ensemble prediction system

• JMA operational Stratospheric assimilated data (Resolution) 1.25º x 1.25º Lon-Lat grid spacing Vertical: 23 Pressure levels (1000hPa - 0.4hPa )

Resolution T106L40 hybrid coordinateTop Boundary 0.4 hPaIntegration Period 34 daysNumber of Ensemble 13 membersPerturbation Method BGM (Breeding of Growing Mode)Initialization Date Every Wednesday and ThursdayInterval of Stored Data Daily (2.5 º x 2.5 º)

Model and Data

*Multiple forecasts from perturbed initial values to give the prediction reliability

Observational features of

3 Major SSWs

(Dec 2001, Jan 2004, Jan 2006)

Observation Latitude-time sections of zonal-mean Temperature and zonal wind at 10hPa

MajorMinorMajorMinorMajor

SSW in Dec 2001 SSW in Jan 2004 SSW in Jan 2006Temperature

90N

EQ

90N

EQ

Zonal wind

Temperature

Zonal wind

Temperature

Zonal wind

Westerlies(blue)

Easterlies(red)

Dec2001

Jan2002

Dec2003

Jan2004

Jan2006

acceleration(blue)

deceleration(red)

Major

Height-time sections of the averaged E-P flux and wave driving over 50-70N

Observation

SSW in Dec 2001 SSW in Jan 2004 SSW in Jan 2006

Fy

Fz

MajorMinorMinorMajor

wave number 1 WN1 WN1

wave number 2+3 WN2+3 WN2+3

3hPa

200hPa

undisturbed, wave number-1 type SSW

disturbed, wave number-1,2,3 contributing SSWs

3hPa

200hPa

Comparison of the Predictabilityof

3 Major SSWs

Forecasts Time evolution of zonal mean temp. of ensemble forecasts at 10hPa, 80N

SSW in Dec 2001 SSW in Jan 2004 SSW in Jan 2006

Major

23,22days before

undisturbed, Wave-1 type LONG predictable period

disturbed, Wave-1,2,3 contributing SHORT predictable period

Weather forecast7 days>

16,15days before 18,17days before

16,15days before 9,8days before 11,10days before

MajorMinor MajorMinor

Dec2001

Jan2002

Dec2003

Jan2004

Jan2006

acceleration(blue)

deceleration(red)

Major

Height-time sections of the averaged E-P flux and wave driving over 50-70N

Observation

SSW in Dec 2001 SSW in Jan 2004 SSW in Jan 2006

Fy

Fz

MajorMinorMinorMajor

wave number 1 WN1 WN1

wave number 2+3 WN2+3 WN2+3

3hPa

200hPa

undisturbed, wave number-1 type SSW

disturbed, wave number-1,2,3 contributing SSWs

3hPa

200hPa

SSW in Dec 2001 SSW in Jan 2004 SSW in Jan 2006

Forecasts Time evolution of normalized Root Mean Square Errors (RMSE) estimated over 20N-90N at 10hPa

Warming peak28 Dec 9 Jan 22 Jan

Forecasts Initialized 12 Dec 1 Jan 12 Jan

21

211

21

1

A

ZZNRMSE

N

iigif

standard deviation

1A

1fZ

1gZ

Forecasts height of Wave number 1 (WN1)

Observational heightof WN 1

Observational amplitude of WN 1

1

0

2

1

0

2

1

0

216days before 8days before 10days before

WN1 WN1 WN1

Dec 2001 Jan 2004 Jan 2006

－ Predictability of wave number 1 component 　－

SSW after the undisturbed situation (Dec 2001)

SSW after the disturbed situation (Jan 2004, Jan 2006)>

SSW in Dec 2001 3-day averaged geopotential height at 500hPa

Observation

SSW in Dec 2001 3-day averaged geopotential height at 500hPa

Ensemble mean ( starting from 16,15 days before the warming peak )

WN1

kk

N

iigif

A

ZZNRMSE

2

2

21

1

standard deviation

kA

fZ

gZ

forecasts height of WN2,3

Observational height of WN2,3

Observational integrated amplitude of WN2,3

SSW in Jan 2004 SSW in Jan 2006

WN2+3

WN1

Forecasts

1

0

2

1

0

2

1

0

2

1

0

2

Predictability of each wave number component

WN 1 > WN 2+3

Time evolution of Normalized RMSE(comparison of WN1 with WN2+3)

8 days before 10days before

WN2+3

Jan 2004 Jan 2006

Jan 2006Warmingpeak9 Jan

ForecastsInitialized

1 Jan22 Jan12 Jan

Jan 2004

Statistics of the Predictabilityof

Recent 12 WarmingsDuring 2001/02 and

2005/06 Winters

Here,

• Extract major and (prominent) minor warmings during 2001/02 and 2005/06 NH winters 12 SSWs

• Classify them into SSWs initiated from disturbed and undisturbed situation

• Estimate their predictable periods

Predictableperiod

Days before the warming peak

0 (warming peak)20days before0

20daysSuccessful

forecasts of SSWpeaks

Failedforecasts

(15days > 5days)

(10days < 15days)10days

15days

15days

5days

Failed Case

SuccessfulCase

t

t

Predictable PeriodDiagram

Results at 10hPa(color) and 100hPa(black)

Undisturbed initials Disturbed initials

12SSWs 　 Major & Minor •For initials near warmingpeaks, predictable periods become short.•SSWs initiated under undisturbed situation show better predictability.

16days

11days

2.5days 3days

Summary

Predictable periods of SSWs differ in a fairly wide range from one to three weeks in terms of RMSE as well as zonal mean temperature.

Predictability is relatively low in case of SSWs with contributions from smaller-scale planetary waves.

Predictability of SSWs initiated under undisturbed situation is better than that under disturbed situation.

10hPa geopotential height

SSW in Dec 2001 SSW in Jan 2004 SSW in Jan 2006

H

LH

L

HH

L

H

*undisturbed, Wave-1 type SSW

in Dec 2001

*disturbed, Wave-1,2,3 contributing

SSWs in Jan 2004 and Jan 2006

4 days 2 days

Observation