Science Review 12-14 May 2015 Boulder, Colorado

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Juliana Dias

Understanding Convective Coupling in Atmospheric Tropical Waves

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Why study convectively coupled tropical waves?

•  tropical waves control local weather;

•  larger scale tropical waves (such as MJO and ER) affect higher latitude weather;

Science Review • Boulder, CO • 12-14 May 2015 2

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Science Review • Boulder, CO • 12-14 May 2015 3

2 days

96 days

Westward & symmetric

modes

Eastward & symmetric

modes

our focus is on synoptic-to-planetary waves that are coupled to rainfall

How to diagnose convectively coupled tropical waves?

Kelvin waves

MJO Equatorial

Rossby waves

TRMM (obs)

Ref:  Jaramillo,  Dias  and  Kiladis  (in  prep.)  

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Science Review • Boulder, CO • 12-14 May 2015 4

2 days

96 days

Westward & symmetric

modes

Eastward & symmetric

modes

Kelvin waves

MJO Equatorial

Rossby waves

TRMM (obs)

How well do models resolve tropical waves?

Ref:  Jaramillo,  Dias  and  Kiladis  (in  prep.)  

GISS CFSv2

ECGEM CAM4

CMIP5 (models)

CMIP5 representation of tropical waves is not uniform

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Science Review • Boulder, CO • 12-14 May 2015 5

2 days

96 days

Westward & symmetric

modes

Eastward & symmetric

modes

CMIP5 representation of tropical waves is not uniform

Kelvin waves

MJO Equatorial

Rossby waves

TRMM (obs) GISS CFSv2

ECGEM CAM4

CMIP5 (models)

Why do models have a difficult time representing tropical waves?

Ref:  Jaramillo,  Dias  and  Kiladis  (in  prep.)  

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

CMIP5 representation of tropical waves is not uniform Science Review • Boulder, CO • 12-14 May 2015 6

2 days

96 days

Westward & symmetric

modes

Eastward & symmetric

modes

Kelvin waves

MJO Equatorial

Rossby waves

TRMM (obs) GISS CFSv2

ECGEM CAM4

CMIP5 (models)

Why do we care about tropical waves?

•  To better understand multiscale tropical-extratropical interactions

•  To improve both weather and climate predictions

Ref:  Jaramillo,  Dias  and  Kiladis  (in  prep.)  

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Science Review • Boulder, CO • 12-14 May 2015 7

(1) How sensitive are synoptic scale waves to cumulus parametrizations and the basic flow?

Ref:  Dias  et  al.,  2013  JAS/  Dias  and  Kiladis,  2014  GRL  

•  Convective coupling induced by Betts-Miller types of cumulus parameterizations have different impacts synoptic scale tropical waves.

lon (°)

time (

days)

(b) MRG div

0 50 100 150 200 250 300 3500

5

10

15

-6 -4 -2 0 2 4 6x 10-3

time

(day

s)

(b)

0 30E 60E 90E 120E 150E 180 150W 120W 90W 60W 30W 00

5

10

15

20

-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06

(a)  KW  amplitude  

(b)  MRG  amplitude  

Time  (days)  

longitude  (deg)  

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(a)

frequ

ency

(cpd

)

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(b)

-10 0 10

200

400

600

800

1000

(c)

pres

sure

(hP

a)

S0°

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(d)

frequ

ency

(cpd

)

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(e)

-10 0 10

200

400

600

800

1000

(f)

pres

sure

(hP

a)

S90° E

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(g)

frequ

ency

(cpd

)

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(h)

-10 0 10

200

400

600

800

1000

(i)

pres

sure

(hP

a)

S180°

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(j)

frequ

ency

(cpd

)

planetary zonal wavenumber-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(k)

planetary zonal wavenumber-10 0 10

200

400

600

800

1000

(l)

pres

sure

(hP

a)

S90° W

U (ms-1

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(a)

frequ

ency

(cpd

)

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(b)

-10 0 10

200

400

600

800

1000

(c)

pres

sure

(hP

a)

S0°

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(d)

frequ

ency

(cpd

)

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(e)

-10 0 10

200

400

600

800

1000

(f)

pres

sure

(hP

a)

S90° E

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(g)

frequ

ency

(cpd

)

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(h)

-10 0 10

200

400

600

800

1000

(i)

pres

sure

(hP

a)

S180°

-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

(j)

frequ

ency

(cpd

)

planetary zonal wavenumber-16 -12 -8 -4 0 4 8 12 16

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8(k)

planetary zonal wavenumber-10 0 10

200

400

600

800

1000

(l)

pres

sure

(hP

a)

S90° W

U (ms-1

Zonal  wind  (ms-­‐1)  Zonal  wave  number  

Freq

uency  (cpd

)  Freq

uency  (cpd

)  

Pressure  (h

Pa)  

Pressure  (h

Pa)  

•  Wave scales are not too sensitive to vertical shear

moister                                                    drier                                                    

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Science Review • Boulder, CO • 12-14 May 2015 8

(2) Are multi-scale interactions key to the existence and propagation of the MJO?

   Ref:  Dias,  Tulich  and  Kiladis,  2011  JAS  /  Dias  et  al.,  2012  GRL    

•  There is no systematic enhancement of synoptic scale variability within the MJO envelope

MJO1                                                                …                                                                                          MJO100  

cycl

es p

er d

ay

MJO events from 1979 to 2012

1.25d

20d

(a)  Eastward  power-­‐spectrum    

(b)  Westward  power-­‐spectrum    

Is  there  a  paJern  of  higher  frequency  disturbances  that  is  common  to  most  MJO  events?  

10/01 10/10 10/20 11/01 11/10 11/20 12/01 12/10 12/20 01/01 01/10 01/20 02/01 02/10 02/20 03/01 03/10 03/200

1

2

3

4

10/01 10/10 10/20 11/01 11/10 11/20 12/01 12/10 12/20 01/01 01/10 01/20 02/01 02/10 02/20 03/01 03/10 03/200

1

2

3

4

(a) DYNAMO

(b) TOGA COAREOMI amplitude

MJO  amplitu

de  

Mme  

MJO  events  

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Science Review • Boulder, CO • 12-14 May 2015 9

(3) What defines the MJO? Zonal wind or Convection?

•  OMI is a useful index for tracking the MJO

convective envelope, including its seasonal changes;

•  OMI produces robust statistics of MJO

primary events; and •  OMI is available at:

http://www.esrl.noaa.gov/psd/mjo/mjoindex/

Ref:  Kiladis  et  al.  2014,  MWR  

•  RMM and OMI are both bivariate MJO

indexes. RMM is primarily a zonal wind based index, and OMI is an OLR-only index.

Phase  2      Indian  Ocean        Phase  3  Phase  2      Indian  Ocean        Phase  3  

RMM (zonal wind) OMI (OLR)

2015  2015  

JAN      FEB      MAR        APR   JAN      FEB      MAR        APR  

Phase  7        Western  Pacific        Phase  6   Phase  7        Western  Pacific        Phase  6  

Phase  1    W

H/Africa        Ph

ase  8  

(a) EOF1 Jan, 15

(b) EOF1 Jul, 15  

 OM

I EO

F pa

tter

ns

N O A A R E S E A R C H • E S R L • P H Y S I C A L S C I E N C E S D I V I S I O N

Summary and Conclusions

Science Review • Boulder, CO • 12-14 May 2015 10

(1)  Both cumulus parameterization and basic flow impact tropical waves in different ways, highlighting differences in the physical processes underlying their convective coupling.

Is that why it is so difficult to represent tropical waves in models?

(2) Because the MJO does not organize higher frequency waves it is possible that model resolution is not critical to the MJO representation; MJO representation in models might be possible in a coarse resolution global model. (3) OLR based indexes are an important target for model development because they better track the MJO convective envelope. A model can have high MJO skill based on a circulation index without properly representing MJO rainfall.