Dynamical MJO Hindcast Experiments: Sensitivity to

Initial Conditions and Air-Sea Coupling

Yehui Chang, Siegfried Schubert, Max Suarez

Global Modeling and Assimilation OfficeNASA/Goddard Space Flight Center

and

Duane Waliser

Jet Propulsion Laboratory, California Institute of Technology

October 21, 2008

Outline

• Measure the capacity of GEOS-5 to predict the MJO.• Focus on the reasons for the temporal changes in

MJO prediction skills.• The experiments were for the period of 1979, Nov

1996 - Aug 1997 and 2002 where there were strong intraseasonal activities.

AGCM Finite-volume dynamical core (S.J. Lin) Moist physics (J. Bacmeister, S. Moorthi and M.

Suarez) Physics integrated under the Earth System

Modeling Framework (ESMF) Generalized vertical coord to 0.01 hPa Catchment land surface model (R. Koster) Prescribed aerosols (P. Colarco) Interactive ozone Prescribed SST, sea-ice

Replay Apply Incremental Analysis Increments (IAU) to reduce shock of data insertion (Bloom et al.) IAU gradually forces the model integration throughout the 6 hour analysis period

GEOS-5 Atmospheric and Coupled Model Replay System

qn

t

total

dynamics(adiabatic) physics(diabatic) q

Model predicted change Correction from DASTotal “observed change”

Analysis

Background (model forecast)Reanalysis (MERRA)

Assimilated analysis(Application of IAU)

03Z 06Z 09Z 12Z 18Z15Z 21Z 00Z 03Z

Initial States for CorrectorReanalysis Tendencies for CorrectorCorrector Segment (1- and 3-hrly products)

CGCM GFDL ocean model (MOM4) A replay of the atmospheric data analysis in the CGCM. Has the potential to massively reduce initialization shocks

Reanalysis MERRA NCEP, JRA-25, ERA

* Scout: coarse resolution, precursor of MERRA system * AGCM: replay Scout data to generate initial states

• Models

AGCM: GEOS-5 at 2X2.5X72

• Replay runs (Scout data) AGCM replay run: 1979, 1996-1997, 2002

• 35-day Hindcasts

AGCM (1034 cases): 1979, Nov1996-Aug1997, 2002

ICs: daily 21z in 1979, 1996-1997, 2002 from replay runs

Dynamical MJO Hindcast Experiments: Sensitivity to Initial Conditions

Analysis tool

• Use Scout 200mb velocity potential from 1979, Nov 1996 – Aug 1997 and 2002.

• Band pass 20-90 days between 30S – 30N.• EOF analysis. Hindcasts and verifications will be

projected to two leading complex EOFs.

* Scout: coarse resolution, precursor of MERRA system * AGCM: replay Scout data to generate initial states

• Models

AGCM: GEOS5 at 2X2.5X72

• Replay runs (Scout data) AGCM replay run: 1979, 1996-1997, 2002 AGCM replay run (w/o Q IAU): 2002

• 35-day Hindcasts

AGCM (1034 cases): 1979, Nov1996-Aug1997, 2002

AGCM ( 365 cases): 2002 (degraded Q in ICs)

ICs: daily 21z in 1979, 1996-1997, 2002 from replay runs

Dynamical MJO Hindcast Experiments: Sensitivity to Initial Conditions

(Q IAU)

• Models AGCM: GEOS-5 at 2X2.5X72 GFDL MOM4 : 1/3x1 in tropics; 1x1 in extratropics; 50 layers Coupling every 30 minutes Sea ice extent taken as observed climatology

• Replay runs (Scout data) AGCM replay run: 1979, 1996-1997, 2002 AGCM replay run (w/o Q IAU): 2002 CGCM replay run: 1979-2005 replay Scout data to initialize ocean and

to generate the initial states

• 35-day Hindcasts

AGCM (1034 cases): 1979, Nov1996-Aug1997, 2002

AGCM ( 365 cases): 2002 (degraded Q ICs)

CGCM ( 669 cases): Nov1996-Aug1997, 2002

ICs: daily 21z in 1979, 1996-1997, 2002 from replay runs

Dynamical MJO Hindcast Experiments: Sensitivity to

Initial Conditions and Air-Sea Coupling

ICs lead to low forecasting skill

May 14, 1979 Apr 3, 1997

Jun 27, 1979 May 3, 1997

Aug 3, 1979 May 6, 2002

Nov 22, 1979

ICs lead to low forecasting skill: cases 1 & 2

ICs lead to low forecasting skill: case 3

ICs lead to high forecasting skill

Apr 19, 1979 Feb 4, 1997 Apr 24, 2002

Jun 8, 1979 May 15,1997 Jun 9, 2002

ICs lead to high forecasting skill: cases 1 & 2

ICs lead to high forecasting skill: case 3

Concluding Remarks:

• There are changes in the MJO forecasting skills between 1979 and 2002. Higher forecasting skills in 2002 appear to be resulting from the better moisture observations from the satellite.

• Both AGCM and CGCM predicted realistic MJO signals. Coupling tends to improve forecasting skills. Interactive air-sea coupling is essential in maintaining the intensity of the MJO in GEOS-5 coupled model.

• GEOS-5 exhibits the different forecasting skills on different phases of the MJO. It has less strength to develop the dry phase of the MJO over the Indian Ocean. Skills are enhanced when intended to develop the wet phase.

• The magnitudes of the surface latent heat fluxes in AGCM are much weaker than those in the coupled system. The phasing of the AGCM respond to the prescribed MJO SST anomalies and the associated surface flux anomalies are incorrectly simulated. This could be due to inadequacies in the model parameterizations. However, by forcing the atmospheric model with prescribed SST, it may be inherently difficult for AGCM to simulate realistic MJO.

• The study suggests that the following additional factors for improving MJO forecasts and simulation may be required: a) accurate initial conditions (e.g. wet and dry phases of an MJO over Indian Ocean); b) coupled to an ocean model to allow for air-sea interactions and realistic surface heat and moisture fluxes; c) improve the model physical package.