The role of the stratosphere in extended-range forecasting

Thomas Jung

Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research

Germany

Overview

Early experiments Relaxation experiments

• Extended-range predictions• Dynamics of two recent cold European winters

o The winter of 2005/06o The winter of 2009/10

Case study: The January 2009 sudden stratospheric warming event

Early experiments: Motivation

Early experiments

F is the mean optimal forcing (adjoint method, 18 cases)

F is applied in the stratosphere only Three 40-day forecasts for each of the winters 1980-

2001• Control• Positive forcing (strong vortex)• Negative forcing (weak vortex)

TL95L60 (cycle 28r1)

Stratospheric response (Z50)

Jung and Barkmeijer (2006); MWR

Stratospheric response (Z1000)

Jung and Barkmeijer (2006), MWR

Downward propagation: ERA-40

Jung and Leutbecher (2007), QJRMS

Downward propagation: TL95L91

Jung and Leutbecher (2007), QJRMS

Conclusions: Part I

“Classical” stratospheric forcing experiments indicate downward control of stratospheric circulation anomalies in the ECMWF model

Downward control is also found in long unforced integrations

The magnitude of the response in unforced experiments looks smaller than in forced experiments

Relaxation experiments

Study the impact of Northern Hemisphere stratosphere on tropospheric forecast skill

Strategy:

• Perform set of control experiments (standard NWP setting)

• Perform additional experiments with forecast error suppressed in the NH stratosphere

• Perform similar experiments for other areas (e.g. tropics)

Experimental setup

TL159L60 (32r1)

88 30-day forecasts (15th of Nov, Dec, Jan and Feb. of the period 1980/81-2000/01)

Initial and boundary conditions from ERA-40 Relaxation towards ERA-40 Persisted SST and sea ice Control, stratospheric and tropical relaxation

experiments

Relaxation regions

Stratospheric relaxation Tropical relaxation

NH Z500 scores

Jung et al (2010a), MWR

NH Z500 forecast error reduction

Jung et al (2010a), MWR

Conclusions: Part II

Relaxation experiments suggest that reducing stratospheric forecast error leads to better tropospheric forecasts

Overall the impacts on Northern Hemisphere forecast skill moderate

Locally larger impacts (e.g. North America and North West Europe)

Results in good agreement with Newman and Sardeshmukh (LIMs)

The cold European winter 2005/06

Jung et al. (2010b), MWR

Seasonal forecasts

Statistical techniques predicted negative NAO Dynamical techniques also predicted negative NAO (ie cold

winter) However, observed circulation anomaly not strictly the NAO

Suggested mechanisms

Folland et al. (2006)• Reemergence of North Atlantic SST anomalies

Scaife and Knight (2008)• Atlantic SST• January SSW

Croci-Maspoli and Davies (2009)• SST and 2mT anomalies in the western North

Atlantic and close to the eastern US seaboard Bader and Latif (2003)

• Cold SST anomalies Indian ocean NAO- Greatbatch and Jung (2007)

• La Nina NAO+

Experimental setup

TL95L60 (32r1) Atmosphere-only with observed SST/sea ice Lagged ensemble (17 members) started in the middle

of November 2005 Calibrations runs with and without relaxation (1990-

2006) Relaxation experiments (various regions)

Z500 anomalies: DJF 2005/06

Z50 anomalies: DJF 2005/06

Z500 anomalies: DJF 2005/06

Ensemble mean anomalies: Polar Z50

Conclusions: Part III

Origin of the cold European 2005/06 has been studied Various mechanisms have been proposed including an

important role of the Northern Hemisphere stratosphere

Our results suggest that the tropics have been important (for the NH troposphere and stratosphere)• Tropical stratosphere: QBO • Tropical troposphere

o South Americao Indian Ocean

The extreme negative NAO winter 2009/10

Jung et al. (2011), GRL

Operational ECMWF forecasts

Jung et al. (2011), GRL

Experimental setup

Jung et al. (2011), GRL

TL159L60 (36R1)

ECMWF Monthly Forecasting System Forecasts started on 1 November 2009 40 Ensemble members Control integration Various sensitivity experiments Hindcasts for each of the configurations

• 1991-2008• 4 ensemble members

Sensitivity experiments I

Jung et al. (2011), GRL

Sensitivity experiments II

D+18-D+32

D+18-D+32

Verifying analysis

Jung et al. (2011), GRL

Conclusions: Part IV

The winter 2009/10 was one of the most negative NAO winters on record

Extremely good monthly and seasonal forecast skill during the second half of the winter

Sensitivity experiments imply that external forcing and the Northern Hemisphere stratosphere played a secondary role

Internal tropospheric dynamics (highly predictable)?

The January 2009 SSW: A case study

Sensitivity experiments have been strongly forced Consider unforced experiments Case selection

• Good extended-range forecasts during the late part of the winter 2008/09

• Strong SSW in January 2009• Did the SSW contribute to the high skill?

Work with Frederic Vitart

Forecasts data

ECMWF VarEPS• T255 up to D+9• T399 from D+10

62 levels in the vertical (up to 5 hPa) Atmosphere-ocean coupling from day-10 50 ensemble members (operationally)

The January 2009 SSW

Start: 8. January 2009 Start: 15. January 2009

Separation into good and poor forecasts

Focus on forecasts started on 15. January 2009 Increase ensemble size to 100 members Separate into best and worst 10 SSW cases

Z500 anomalies

Z500 forecast anomalies (initialized on 15. January 2009)

The role of vertical resolution

Start: 8. January 2009 Start: 15. January 2009

Conclusions: Part V

Major SSW occurred in January 2009 Extended-range forecasts for Europe have been quite

skilful following the SSW Role of the SSW has been assessed by splitting

ensemble into two sub-groups (good and poor SSW forecasts)

Impact of the SSW seem to have played a secondary role

More cases need to be considered in future studies

Some general thoughts

Stratosphere provides a source of extended-range forecast skill in the Northern Hemisphere troposphere

Some areas might benefit in particular (e.g. Europe) However, the influence is likely to be moderate at best

(disagreement with Douville et al?) It is somewhat surprising that there is such a spread

regarding possible mechanisms of recent cold European winters

Further reading: