23
Simulations of the Madden- Julian Oscillation by Global Models: Current Status Chidong Zhang, Min Dong RSMAS, University of Miami Harry Hendon, Andrew Marshall BMRC Eric Maloney Oregon State University Kenneth Sperber PCMDI, Lawrence Livermore National Laboratory Wanqiu Wang CPC/NCEP

Simulations of the Madden-Julian Oscillation by Global Models: Current Status

Embed Size (px)

DESCRIPTION

Simulations of the Madden-Julian Oscillation by Global Models: Current Status. Chidong Zhang, Min Dong RSMAS, University of Miami Harry Hendon, Andrew Marshall BMRC Eric Maloney Oregon State University Kenneth Sperber PCMDI, Lawrence Livermore National Laboratory Wanqiu Wang CPC/NCEP. - PowerPoint PPT Presentation

Citation preview

Page 1: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Simulations of the Madden-Julian Oscillation by Global Models:

Current Status

Chidong Zhang, Min DongRSMAS, University of Miami

Harry Hendon, Andrew MarshallBMRC

Eric MaloneyOregon State University

Kenneth SperberPCMDI, Lawrence Livermore National Laboratory

Wanqiu WangCPC/NCEP

Page 2: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Objectives:

(1) To evaluate how well we currently can simulate the MJO

using global climate and weather forecast models

(2) To gain insight into MJO dynamics from the success and

failure of global model simulations

Issues: - What is the improvement during the last decade?

- What are the remaining common problems?

- How does air-sea coupling affect MJO simulations?

- How does the mean background state affect MJO simulations?

Page 3: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Models

Uncoupled

AGCM

Coupled

CGCM

Institute

BAM3(20 yrs)

BAM3C(20 yrs)

BMRC

GFS03(20 yrs)

GFS03C (CFS)(20 yrs)

NCEP

CAM2R(16 yrs)

CAM2RC(15 yrs)

NCAR/OSU

ECHAM4 (20 yrs)

ECHO-G(20 yrs)

MPI

Page 4: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Atmosphere ModelsModels

(Institute)

Horizontal Resolution

Vertical Levels

(top level)

Cumulus Parameterization Integration

BAM3

(BMRC)

T47 (2.5˚)

17

(10 hPa)

Mass flux (Tiedtke 1989)

Adjustment closure (Nordeng 1994)

1982 - 2001

GFS03 (NCEP)

T62

(1.8˚)

64

(0.2 hPa)

Mass flux

(Hong and Pan 1998)

1979 - 1998

CAM2R(NCAR/OSU)

T42

(2.8˚)

26

(3.5 hPa)

Relaxed Arakawa-Schubert

(Moorthi and Suarez 1992)

16 years

ECHAM4

(MPI)

T42

(2.8˚)

19

(10 hPa)

Mass flux (Tiedtke 1989)

Adjustment closure (Nordeng 1994)

20 years

Page 5: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Ocean Models

Coupled Run Ocean Models Meridional Resolution

Zonal Resolution

Vertical Levels Flux Correction

BAM3C ACOM2 0.5˚in 9˚N-9˚S

1.5˚ near the poles

2˚ 25

(12 in upper 185m)

Yes

GFS03C MOM3 1/3˚ in 10˚N-10˚S

1˚ beyond 30˚N/S

1˚ 40

(27 in upper 400m)

No

CAM2RC SOM 0 0 1 Yes

ECHO-G HOPE-G 0.5˚ in 10˚N-10˚S

2.8˚beyond 30˚N/S

2.8˚ 20

(8 in upper 200m)

Yes

Page 6: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Observations: (1) NCEP/NCAR reanalysis zonal wind at 850 hPa (U850) (Kalnay et al

1996)

(2) CMAP precipitation (Xie and Arkin 1997)

Analysis Method: (1) Time-space spectrum (Hayashi 1979) of unfiltered data

(2) MJO reconstruction using Hilbert SVD (Zhang and Hendon

1997) applied to intraseasonally (20-90 day) band-passed data

(3) Seasonal cycle and geographic distribution of the MJO (Zhang

and Dong 2004)

Page 7: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

U850 precipitation

observations

BAM3

BAM3C

GFS03

GFS03C

CAM2R

CAM2RC

ECHAM4

ECHO-G

Time-space spectra

10˚N-10˚S/60-180˚E

Eastward power > westward power

Wind signal stronger than precipitation

Air-sea interaction enhance eastward power

Page 8: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

OBS BAM3 BAM3C GFS03 GFS03C CAM2R CAM2RC ECHAM4 ECHO-G

U850 3.5 2.2 2.8 2.7 4.6 2.0 3.2 2.0 2.7

P 2.4 1.2 1.3 1.2 1.7 1.5 1.9 1.3 1.2

Ratio of eastward vs. westward intraseasonal power for 850 hPa zonal wind (U850) and precipitation (P). Intraseasonal power is defined as within the window of 30 - 90 days at zonal wavenumber 1 for U850 and zonal wavenumbers 1 and 2 for P.

PEastward/PWestward

Simulated signal in wind is more realistic than simulated signal in precipitation.

Air-sea interaction helps strengthen the signals for all models except for precipitation in ECHO-G.

Page 9: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Number of Leading HSVD Modes for MJO Reconstructionand Accumulative Fractional Variance

OBS BAM3 BAM3C GFS03 GFS03C CAM2R CAM2RC ECHAM4 ECHO-G

U850 4

(49%)

10

(58%)

4

(40%)

4

(44%)

4

(47%)

2

(22%)

6

(48%)

4

(36%)

4

(41%)

P 4

(36%)

4

(20%)

6

(30%)

4

(23%)

2

(14%)

4

(25%)

4

(47%)

6

(30%)

6

(31%)

Isolation of the MJO signal

Only outstanding Modes are used (Based on the Rule of North et al 1982)

Page 10: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

ECHO-G

GFS03

CAM2R CAM2RC

GFS03C

BAM3

Obs

BAM3C

ECHAM4

Lag-regression upon MJO of U850 at 160˚E and 0˚N

Equatorial U850 (contours) Equatorial precipitation (colors).

Propagation of the MJO

Page 11: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

(a) OBS

(b) BAM3

(d) GFS03

(e) GFS03C

(g) CAM2RC

(f) CAM2R

(c) BAM3C

(i) ECHO-G

(h) ECHAM4

Zero-lag regression upon MJO U850 at 160˚E and 0˚N.

U850 (vectors)precipitation (colors)

Horizontal Structure

Page 12: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Contours: MJO varianceColors: Mean

(a) OBS

(b) BAM3

(d) GFS03

(e) GFS03C

(f) CAM2R

(c) BAM3C

(i) ECHO-G

(h) ECHAM4

(g) CAM2RC

U850 Geographic distribution

December -March

Page 13: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

(a) OBS

(b) BAM3

(d) GFS03

(e) GFS03C

(f) CAM2R

(c) BAM3C

(i) ECHO-G

(h) ECHAM4

(g) CAM2RC

Contours: MJO varianceShadings: Mean

Precipitation Geographic distribution

December -March

Page 14: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

BAM3 BAM3C GFS03 GFS03C CAM2R CAM2RC ECHAM4 ECHO-G

U850 1.5 1.1 0.8 1.0 1.4 2.4 0.9 1.5

P 0.4 0.6 0.4 0.1 0.5 0.8 0.6 1.1

Modeled Variance / Observed Variance

December – March

(15˚S- 15˚N, 50 - 180˚E)

Page 15: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Contour: MJO Variance

Color: Mean

60E - 180˚E average

U850 Precipitation U850 Precipitation

OBS

BAM3

GFS03 GFS03C

CAM2R

BAM3C

ECHO-GECHAM4

CAM2RC

OBSMJO Seasonal migration

Page 16: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

RMS MJO: 15˚S - 15˚N, 50 - 180˚E RMS mean: 15˚S - 15˚N, 50 - 270˚E

MJO RMS error vs Mean state RMS error

(December - March)

Blue: Uncoupled Red: Coupled

Effect of mean state

Page 17: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Mean variance of MJO precipitation (contour) overlaid with mean moisture convergence

December - March

850 hPa MC

925 hPa MC

Effect of mean state

Page 18: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Summary

Improvement:

(1) intraseasonal, planetary-scale, eastward propagating spectral power in winds

stronger than westward propagating spectral power;

(2) realistic eastward phase speed of the MJO in the western Pacific.

Common problems:

(1) weak MJO signal in precipitation,

(2) unrealistic phase relation between precipitation and wind (maximum

precipitation not in low-level westerlies in the western Pacific),

(3) split of precipitation maxima in the western Pacific,

(4) seasonal migration unrealistic in many models.

Page 19: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Summary (cont.)

Important issues:

(1) Effects of air-sea coupling on MJO simulation are highly model-

dependent.

(2) Biases in MJO simulations are related to biases in simulated mean low-

level zonal wind and mean precipitation.

(3) The MJO activity depend on mean boundary-layer (925 hPa) moisture

convergence.

(4) The incoherence between MJO wind and precipitation in the simulations

raises questions regarding our understanding of the MJO dynamics.

Page 20: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Thank You!

Page 21: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Time - latitude plot of variance in MJO U850 (contour, interval of 2 m2 s-2) and precipitation (contour, interval of 2 mm2 day-2) averaged over 60 - 180˚E. Mean U850 (color, m s-1, zero outlined by white contours) is overlaid with MJO U850 and mean precipitation (color, mm day-1) overlaid with MJO precipitation.

Page 22: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Scatter diagrams of RMS differences between individual simulations and observations in (a) MJO U850 variance (m2 s-2) and mean U850 (m s-1), (b) MJO precipitation variance (mm2 d-2) and mean precipitation (mm d-1), (c) MJO precipitation variance (mm2 d-2) and mean 925 hPa moisture convergence (g kg-1 m-1),and (d) MJO precipitation and mean 850 hPa moisture convergence. Symbols represent: circles for BAM3/BAM3C, crosses for GFS03/GFS03C, plus signs for CAM2R/CAM2RC, and squares for ECHAM4/ECHO-G, with blue for uncoupled and red for coupled simulations. RMS differences were calculated over 15˚S - 15˚N, 50 - 180˚E for the MJO variables and 15˚S - 15˚N, 50 - 270˚E for the mean state variables during December - March. Arrows in (d) highlight changes from uncoupled to coupled simulations.

(a) (b)

(c) (d)

Page 23: Simulations of the Madden-Julian Oscillation by Global Models:  Current Status

Mean variance of MJO precipitation (contour) overlaid with mean moisture convergence (g kg-1 s-1) at (a) 850 hPa and (b) 925 hPa for December - March. Contour intervals are 2 mm d-1 starting from 1.

850 hPa MC

925 hPa MC