Boreal Summer Intraseasonal Oscillation (BSISO)

Bin WangDepartment of Meteorology and

IPRC, University of Hawaii

1.MJO and BSISO2.Observed characteristics of BSISO3.Mechanisms for BSISO4.Role of atmosphere-ocean interaction5.Impacts of BSISO6.Modeling, Prediction and Predictability

I. MJO and BSISO

boreal winter (MJO)

boreal summer (BSISO)

Wang and Rui 1990

How ISO behavior is affected by annual cycle?

EOF1 (10%) EOF2 (10%)

EOF1 (12%) EOF2 (8%)

Principal Modes of

ISO (30‐60 day) during DJF

and  JJA, 1979‐2010 

DJF

JJA

Do we need a bimodal representation of  Tropical ISO?

MJO in N.H. Winter MJO in N.H. Summer

From MJO working group website

Propagation

Kikuchi, Wang, Kajikawa, 2011 Climate Dyn

Bimodal

representation of the  tropical ISO:

MJO and BSISO

Method

25-90 day filtered OLR

JJA DJFEEOF (0,5,10 day)

BSISO mode

Projection: 25-90 day filtered OLR

BSISO entire time series

MJO entire time series

MJO mode

BSISOEEOF2 EEOF1

MJOEEOF1 EEOF2

Bimodal ISO modes: MJO and BSISO

PCs: MJO, BSISO & WH04

MJO

BSISO

WH04

||OLRBSISO

||

||OL

RM

JO||

MJO vs BSISO

Phase space and category

MJO BSISO

PC1

PC2

phase 1

phase 2 phase 3

phase 6phase 7

phase 8

eastern Indian Ocean

Afr

ica

& w

eter

n IO

Mar

itim

e C

ontin

ent

phase 4

Western Pacific

Bay of Bengalea

ster

n N

P &

equ

ator

ial I

O

Indn

ia

& M

ariti

me

Con

tinen

t

western North Pacific

phase 5

PC2

PC1

Composite Life CycleM

JOB

SISO

Fractional varianceD

JFJJ

AMJO

BSISO

WH04

WH04

month

Ave

rage

num

ber o

f da

ys p

er m

onth

Seasonal variation of the two ISO modes

BSISO

MJO

2. Observed Characteristics of BSISO

OLR

1979~2009

1979~2010

(W/m2)2

1979-20010

Camball-Cook and Wang 2001

Variance distribution Propagation

MJ

Jul

ASO

MJJ

ASO

BSISV: Discovery of northward propagation

•

Yasunari (1979, J. Meteorol. Soc. Japan, 57, 227-242)•

Yasunari (1980, J. Meteorol. Soc. Japan, 58, 225-229)

•

Sikka and Gadgil (1980, MWR, 108, 1840-1853)

Latit

ude

Latit

ude

Time-latitude plots of the location and width of the maximum cloud zone at 90oE (after Sikka and Gadgil 1980)

Time-latitude plots of the location and width of the maximum cloud zone at 90oE (after Sikka and Gadgil 1980)

(a) TMI rain rate

(75- 100E,5S-5N)

20-5

0day

ano

mal

ous r

ain

rate

(red

das

hed,

m

m/d

ay)

Dai

ly ra

in ra

te (b

lue,

mm

/day

)

26 37 26 40

39 43 28

31 32 49 32

36 32 24 30

33 34 26 27 32 29

38 28 28

30 33 37 29

(b) 20-50day 3B42 rain rate (75-100E)

1998

1999

2000

2001

2002

2003

2004

BSISO in EIO and N-propagation

Wang et al. 2005)

Composite life cycle of BSISO rain rate (contour) & SST (shading)

1 2 3

3

4

4

1

1

1 2 3

3

4

4

1

1

Schematic structure/movement of BSISO

Wang, Kikuchi and Webster 2005

Phase1: genesis in western EIO; Phase 2: IntensificationPhase 3: Bifurcation, formation of tilted rain bandPhase 4:Northeastward propagation

30-60 day correlation: WNPSMI Quai-Biweekly correlation WNPSMI

Observed Characteristics of BSISO

1. Action centers of convection shifts to monsoon trough regions (Kemball-Cook and Wang 2001)

2. Initiation in the western Equatorial Indian Ocean (60-70E) (Wang, Webster, Teng

2005)

3. NW-SE tilted rain band (Farantii

et al. 1997, Annamalai and Slingo

2001, Kemball-Cook and Wang 2001, Waliser

et

al. 2003) The equatorial eastward propagating MJO tends to The equatorial eastward propagating MJO tends to bifurcatebifurcate

polewardpoleward

near Sumatra (Maloney and Hartmann near Sumatra (Maloney and Hartmann

1998, 1998, KemballKemball--Cook and Wang 2001, Lawrence and Cook and Wang 2001, Lawrence and Webster 2002). Webster 2002).

4. Northward propagation in the Bay of Bengal (Yasunari 1979, 1980, Sikka

and Gadgel

1980, Krishnamurti

and

Subrahmanyam

1982 among others) and northwestward propagation in the WNP (Nitta 1987)

3. Physical Mechanism of BSISO

How are the active/break cycles of ISM re-initiated or MISO

maintained?

How are the active/break cycles of ISM re-initiated or MISO maintained?

Hypotheses1. Circumglobal propagation of the upper-level divergent waves of MJO (Julian and Madden 1981, Lau and Chan 1986 among many others)2. Midlatitude forcing: Forced by midlatitude

Rossby

wave train

(Hsu et al. 1990) or by injection of PV from SH (Rodwell

1997)3. Forcing from decaying off-equatorial Rosbby waves in IM region re-initiates of equatorial convective anomalies by decaying off-

equatorial Rossby

waves in ISM region (Wang and Xie

1997)4. Feedback between hydrological processes in the atmosphere and radiation processes (Hu

and Randal 1994, Stephens et al. 2004)

5. Self-induction mechanism (Wang, Webster, Teng

2005)

200 hPa

velocity potential has a salient stationary component

phase 1

phase 2

phase 3

phase 4

Upper- level

divergence waves may not be essential for re-

initiation

Circum-global navigation?

:sfc. div. & rainfall:sst & rainfall

Lag

corr

elat

ion

coef

ficie

nt

Lag (days)

Genesis in Central EIO: 5S-5N, 60-70E

Surface convergence leads the genesis by 3-4 days

SS warming leads genesis by 6-7 days.

Low-level convergence and local warming may be important

Re-initiation ScenarioRainfall (contours) and SST (color) Surface winds and divergence

Wang, Webster, Teng

2005

The in situ surface wind convergence and sea surface warming that initiate new rainfall anomalies result from the forcing of the previous active monsoon, suggesting a self-Induction mechanism to sustaine BSISO.

Why does the summer monsoon ISO have a NW-SE slanted structure? How does the tilted rain band form?

Can we simulate this structure?

BSISO Model

The two and half layer model includingObserved Mean flows (U,V,W,T)Realistic qs

or SST

Nonlinear heating (SST dependent trigger function and positive only heating)

Initial value problem

(1)

(2)

(3)

(4)

(5)

vvMy

yuxv

)(

0

pyv

xu

ppMpQ

pCRpSpt c

p

)()(

s

u

s

u

s

u

p

p v

p

p cc

p

pqME

gdpQ

Lgdp

yqv

xqu

gdpqt )(1

uuMx

yutu

)(

Model Equations on equatorial beta-plane

11

)(

HH

ss p

pqpq )( SSTqq ss Wang 1988

Mean Flow Terms

pu

pu

yuv

yuv

xuu

xuuuM ''

''

' ')(

pv

pv

yvv

yvv

xvu

xvuvM ''

''

' ')(

pyv

pyv

pxu

pxu

pM

2

'22

'2

s

u

p

p gdp

yqv

xquqM )(

Structure of three-layer model of ISO

mbPs 1000

mbPe 900

mbP 5002

mbP 1000

mbP 3001

mbP 7003

2

00

111 ,, vu

333 ,, vu

b

bb vu ,PBL

p

p

eq

3q

1q

BSISO Model

The two and half layer model includingObserved Mean flows (U,V,W,T)Realistic qs

or SST

Nonlinear heating (SST dependent trigger function and positive only heating)

Initial value problem

July mean state (ER40)

(a)

(b)

(c)

(e)

(d)

(f)

(g)

EQ20S

20N

EQ20S

20N

EQ20S

20N

EQ20S

20N

EQ20S

20N

EQ20S

20N

EQ20S

20N

0

10

6

2

14

18

22

26

Mean flows and SST distribution trap ISO in eastern hemisphere

Simulated boreal summer convectively coupled Kelvin-Rossby waves

Northward propagation component

In the model, the NW-SE slanted precipitation anomalies in the monsoon regions forms due to emanation of the moist Rossby

waves from the equatorial rainfall anomalies over the maritime continent.

Interaction between moist Rossby

wave

and the vertical shear of the mean monsoon provides a mechanism for the formation of the slanted ISO rain band.

Mean flows removedUniform SST

Only Monsoon vertical Shear included Drbohlav and Wang 2005

What give rise to the northward propagation of ISO over the summer monsoon regions?

Northward and westward propagation mechanism in monsoon regions

Northward Propagation•

The land surface heat fluxes into the boundary layer can destabilize the atmosphere ahead of ascending zone, causing a northward shift of the convection zone (Webster 1983)

•

The continuous northwestward emanation of Rossby waves from the equatorial Kelvin-Rossby

wave packet

when the latter pass through the maritime continent (Wang and Xie

1997)

•

Easterly vertical shear, boundary layer advection and air sea interaction (Drbohlav

and Wang 2004, Jiang

et.

al.2004)Westward Propagation

•

Unstable baroclinic

waves (Lau and Peng1990) •

Equatorial Rossby

waves destabilized by easterly vertical

shear and interactive convective heating (Xie

and Wang 1996)

Webster et al. 1983

Wang and Xie 1997Lawrence and Webster 2001

Review northward propagation of boreal summer ISO

Jiang et al. 2004; Drbohlav and Wang 2005Interaction between vertical shear and convection

Fu et al. 2003Air-sea interaction

The land surface heat fluxes into the boundary layer can destabilize the atmosphere ahead of ascending zone, causing a northward shift of the convection zone

Northward propagation is a component of eastward movement of the slated rain band

ω( 0.04 Pa/s) and relative velocity (×10-6s-1)

phase 4

phase 5

phase 6

phase 7

Observation: Barotropic

vorticity

leads convective anomalies in N-propagation.

ECHAM Model: Vorticity

leads convection anomaliesvertical velocity vorticity geopotential

height

divergence specific humidity temperature

Jiang et al. 2003

y

(North)

x (East)

3u CONV

z

u

1u

PBL

0

CONV

'w

AAn atmospheric internal dynamic mechanism for northward propagatin atmospheric internal dynamic mechanism for northward propagation: monsoon on: monsoon easterly vertical shear provides a easterly vertical shear provides a vorticityvorticity

source, which, upon being twisted by the source, which, upon being twisted by the

northnorth--south varying vertical motion field associated with the south varying vertical motion field associated with the RossbyRossby

waves, waves, generates generates positive positive vorticityvorticity

north of the convection, creating boundary layer north of the convection, creating boundary layer

moisture convergence that favor northward movement of the enhancmoisture convergence that favor northward movement of the enhanced rainfall. ed rainfall.

yUv

t T

How easterly vertical shear pulls the RW Northward

4. Roles of Atmosphere-Ocean Interaction in BSISO

Findings•A-O interaction enhances ISO variability (Flateu

1997, Wang and Xie

1998, Waliser

et al. 1999,…)•AGCM (AMIP run) failed to simulate correct SST-Precipitation relationship: in phase in the AGCM models but 90 degrees out of phase in reality. (Wu et al. 2002)•CGCM and AGCM alone yield fundamentally different ISO solution, coupling leads to realistic SST-precipitation relationship (Fu et al. 2003).•Coupling between atmosphere and ocean add predictability to boreal summer ISO (Fu et al. 2006)

Questions•How does ocean intraseasonal

variability feedback to atmospheric ISO?

•What are precise relationships between the SST and surface heat fluxes?•What are the relative roles of entrainment, upwelling, and advection in controlling SST ISV? To what extent theses processes are dependent of atmospheric forcing? (or does ocean processes add noise to ISO?)

Kemball-Cook and Wang, (2001)

Observed Characteristics of AOI in BSISO

The local SST-rainfall phase relationship differs between the equatorial regions and off-equatorial monsoon regions.

Observed Relationship between SST and rainfall anomalies

Does air-sea coupling impact the northward propagation of monsoon intraseasonal

oscillations? If so, how?

CMAP Rainfall

Coupled

Daily Forced

Mean Forced

Roles of air-sea coupling in northward propagation

(Fu and Wang 2004)

Phase Relationships between Rainfall and SST

Arabian Sea

Bay of Bengal

Fu and Wang, (2004)

How AOI Intensifies Northward Propagating ISO?

Surface Wind &Latent Heat Flux

Cloud & Solar Radiation

Mixed-layer Depth

D

D

D

Propagating Air‐sea interaction mechanism

Fu et al. (2003), Wang et al. (2009)

BSISO over the Philippine Sea

How AOI affects BSISO over the Philippine Sea

Wang and Zhang 2002

5. Impacts of BSISO on Monsoon, extreme and midlatitude

Waliser

2005

BSISO is a dominant mode of Monsoon ISO

Monitoring BSISO Using monsoon circulation indices

Wang, Wu, Lau 2001, JC

Asian summer monsoon ISO:

Time series of Indian Summer Monsoon Index (left)Western

North Pacific Summer Monsoon Index (right).(top) Mean 365-day annual cycle (lower three panels) The thin lines are daily anomaly values, the thick lines are 30-60 day band-passed values for the years 1979, 1992 and 2008.

ISO of Monsoon circulation

ISM EA-WPSM

Origin of Synoptic-Scale Wave Train (SWT) in WNP

Lau and Lau (1990) :An alternative positive and negative vorticity

wave train

withtimescale: 2-8 days,wavelength: 2500 km,propagation: northwestward.

Questions:What is the origin of the synoptic wave train? What determines its zonal wave- length and phase propagation?

(c)

Figure 1 (a) Enhanced and (b) relaxed WNP 

monsoon trough during active and break 

phases of ISO in Peak TC season from July to 

October. Shading denotes 850 hPa

vorticity

and the dots represent locations of the first 

RI reported. The green curves represent the 

location of the monsoon trough. The dashed 

box highlights regions in which the largest 

differences between dry and wet phases of 

intraseasonal

oscillation (ISO) are observed. 

(c) The number of RI in the dashed box (10‐

20 ºN, 110‐150 ºE) as function of ISO in the 

composite nine phases of ISO. The 

composition was made form 55 cases during 

June through October. Phases 1 and 5 

represent, respectively, the strongest and 

weakest convection phases over the 

equatorial western Pacific (EWP, 5oS‐5oN, 

140oE‐160oE). Thus, the maximum 

occurrence of RI over the WNP lags the peak 

wet phase of the EWP ISO by three phases 

(about 12 days) and the minimum 

occurrence of RI over the WNP lags the peak 

wet phase of the WNP ISO by seven phases 

(about 28 days).

66

MJO Impacts –

Tropical Cyclones

The MJO often leads to “bursts”

and

“lulls”

in tropical cyclone activity when active

Periods generally last for about two weeks

Due to changes in both lower and upper-level winds and convection

H H H

H

L

L

L

L

MJO and the Record-Breaking East Coast Snowstorms in 2009/10

Bar: Eastern US snowLine: Central Pacific MJO

Fig. 8. The schematic diagrams illustrating the typical patterns of MJO-

teleconnection

at Phase 3 Contour : rotational wind at 300 hPa. Letter A (C) :anticyclonic

(cyclonic) circulation anomaly

The winter mean SST anomaly during El Niño and La Niña is shaded in red (T ≥

0.5°C) and blue (T ≤

-0.5°C), respectively.

The dotted blue (red) arrow denotes cold (warm) advection. The yellow(green) rectangle denotes dry(wet) precipitable

water.

Cloud shape in green represents convection associated with MJO and the cloud shape in orange denotes the subsidence at Phase 3.

Green thick line indicates the succession path of cyclonic and anticyclonic

anomalies departed from the convection.

El Nino

La Nina

Fig. 9. The schematic diagrams illustrating the typical patterns of MJO-

teleconnection

at Phase 7 Contour : rotational wind at 300 hPa. Letter A (C) :anticyclonic

(cyclonic) circulation anomaly

The winter mean SST anomaly during El Niño and La Niña is shaded in red (T ≥

0.5°C) and blue (T ≤

-0.5°C), respectively.

The dotted blue (red) arrow denotes cold (warm) advection. The yellow(green) rectangle denotes dry(wet) precipitable

water.

Cloud shape in green represents convection associated with MJO and the cloud shape in orange denotes the subsidence at Phase 3.

Green thick line indicates the succession path of cyclonic and anticyclonic

anomalies departed from the convection.

El Nino

La Nina

L

L

L

HH

L LL

HHH

H HH

HL

L

Possible positive feedback between the Eurasian wavetrain and ISM on intraseasonal timescale

The strong convection over northern ISM region is initially triggered by the anomalous central Asian high within the wavetrain extending from the northeastern Atlantic to East Asia

The northeastern Atlantic anomalous high is presumably excited by efficient kinetic energy extraction from the basic state (Simmons et al 1983).

Barotropic instability

Vertical shear effect

The increased easterly vertical shear would increase equatorial Rossby wave instability of the atmosphere by providing wave available potential energy, thereby increasing monsoon precipitation (Wang and Xie 1996, Xie and Wang 1996).

The convection in turn excite a Rossby-wave response to reinforce the central Asian high and downstream circulation anomalies of the wavetrain, through Rossby wave dispersion.The baroclinic structure of adjacent circulation associated with the ISM convection indicates the diabatic heating effect of the ISM.

“Monsoon-desert” mechanism

Rossby wave dispersionAnd wave guide effect

the westward retreat of the central Asian high after its establishment and the outbreak of convection over the ISM region can be regarded as a Rossby wave response to the diabatic heating

6. GCM modeling, Prediction and Predictability of BSISO

Progress

BSISV in GCMs has received much less attention compared to the simulation of boreal winter MJO–AGCMs: Fennessy

and Shukla

(1994), Ferranti

et al. (1997), Sperber

et al. (2001), Waliser

et al. (2003)

–CGCMs: Kemball-Cook et al. (2002), Fu et al. (2003), Fu and Wang (2004), Rajendran

and

Kitoh

(2006), Sperber

and Annamalai

(2008)

Modeling•

What is the correct heating partitioning between the convective and stable precipitation?

•

What is the correct heating partitioning between the small-scale high frequency and large-scale, low frequency disturbances?

•

What is typical heating profile for synoptic and ISO time scale?

•

What are the typical vertical profile of specific humidity and moist static energy?

•

What is the precise relationship between anomalous SST, BL moist static energy, convective instability, the boundary layer and lower tropospheric

moisture

convergence, and precipitation associated with ISO?

Fig. 17

Common problems:1.

EIO activity center

2.

Northward pathway in Bay of Bengal

3.

Northwest pathway in the western North Pacific

Common strengths1.

Weakening over the MC

2.

Off equatorial activity centers

SD of 20-90 day filtered rainfall (mm/day) for May-Oct from the CMAP for 1979 to 1998 and for the ten AGCMs (lower). In the case of the models, there were 20 summer seasons of data, i.e. ten members each consisting of two years. From Waliser et al. (2003).

The tilted rainband

•

CLIVAR AAM experiments, 1997/98; 10 member ensembles; weekly SST prescribed

•

Typically, AGCMs poorly represent the BSISV tilted rainband (Waliser et al. 2003, Clim. Dynam., 21, 423-446)

Figures kindly provided by D. Waliser

Figures kindly provided by D. Waliser

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

4 0E 6 0 E 8 0 E1 0 0E1 2 0 E14 0 E1 60 E 1 803 0 S

1 0 S

1 0 N

3 0 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0E 6 0 E 80 E10 0 E1 20 E1 4 0E16 0 E 18 030 S

10 S

10 N

30 N

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

4 0 E 60 E 8 0E1 00 E1 2 0E1 40 E1 6 0E 1 8 030 S

10 S

10 N

30 N

PC4 regression with 20-100 day filtered OLR (Wm-2; best fit to AVHRR Day 10 CsEOF

using pattern correlation)

•

Compared to the GCMs analyzed by Waliser et al. (2003) the newer coupled models are better at representing the BSISV, but…

a) AVHRRa) AVHRR

c) ECHO-Gc) ECHO-G

e) CGCM3.1 (T47) e) CGCM3.1 (T47)

f) CNRM-CM3 f) CNRM-CM3 i) GFDL-CM2.0 i) GFDL-CM2.0

j) GFDL-CM2.1 j) GFDL-CM2.1

k) GISS-AOM k) GISS-AOM

l) IPSL-CM4 l) IPSL-CM4

m) MIROC3_2_hires m) MIROC3_2_hires

n) MIROC3_2_medres n) MIROC3_2_medres

h) ECHAM5/MPI-

OM

h) ECHAM5/MPI-

OM

o) MRI-CGCM2.3.2 o) MRI-CGCM2.3.2

d) CCSM3.0 d) CCSM3.0

b) ECHAM4/OPYCb) ECHAM4/OPYC

g) CSIRO Mk3 g) CSIRO Mk3

1111

99

1212 1414

1212 99

1515

99

Miura et al. (2007).

3B42

MJO simulated by NICAMNICAM 7 km

NICAM 850 vorticity

BSISO Prediction

•What is physical basis for intraseasonal predictability?

•

To what extent the is ISV predictable?•

Improvement of climate models: Cloud resolving models? Multi-scale interaction?

•

Techniques for better initialize coupled models

81

Statistical Predictions –

Jones et al. 2004

• The model uses PCs of EOF analysis of 20-90 day OLR anomalies•

Forecast models use information from 5 most recent pentads to predict future PCs patial

structures are obtained by reconstructing the fields of OLR using the forecasts PCs and the associated EOFs

Forecast Skill

Real-time Multivariate MJO (RMM) index (Wheeler and Hendon04)Observed RMM

Forecasted RMM Eigenvector

Project the combinedeigenvector to predicted variables

(daily OLR, U850,U200)

Forecasted PC1, PC2(RMM index)

Verification method: RMM index

Statistical Predictions –

WH Lagged Linear Regression

Forecast Skill

From Gottschalck

84

RMM1 and RMM2 values for the most recent 40 days and forecasts from the ensemble Global Forecast System (GEFS) for the next 15 days

light gray shading: 90% of members

dark gray shading: 50% of forecasts

Yellow Lines – 20 Individual Members Green Line – Ensemble Mean

Dynamical Predictions –

GEFS

From Gottschalck

85

GEFS MJO Index Forecast Skill

--Ensemble Global Forecast System--Horizontal resolution is T126--21 member ensemble

--Includes all days available(all seasons, non-MJO days so skill is somewhat low)

--In some cases, the GEFS has been able to provide information for MJO initiation and demise

--Only 2.5 years of data available

From Gottschalck

Better Initial Conditions

Boost Intraseasonal

Forecast SkillFu and Wang et al. (GRL, 2009)

The amplitude of convective activities in the NCEP reanalysis (left panel, bottom) was found to be less than observed 

(left 

panel: 

top, 

TRMM 

3B42; 

left 

middle: 

GPCP) 

by 

a 

factor 

of 

two 

to 

three. 

Three 

forecasting 

experiments 

were 

conducted to explore the impact of strengthening the signal in the NCEP initial conditions on forecasting 

aspects 

of 

the 

monsoon 

intraseasonal

oscillation 

(right 

panel).   With 

the 

original 

NCEP 

reanalysis 

as 

initial 

condition, 

the 

850‐

hPa 

zonal 

winds 

and 

rainfall 

are 

predictable 

with 

some 

skill 

only 

about 

a 

week 

in 

advance 

over 

the 

global 

tropics 

(30S–30N) and Southeast Asia (10N–30N,60E–120E). Predictability increases steadily with increasing the amplitudes 

to 

2 

times 

and 

3 

times 

the 

NCEP 

initial 

conditions.   When 

the 

signals 

in 

initial 

conditions 

are 

recovered 

to 

a 

level 

similar 

to 

that 

in 

the 

observations 

(the 

last 

experiment), 

monsoon 

forecast 

skill 

reaches 

25 

days 

for 

850‐hPa 

zonal 

winds and 15 days for rainfall over both the global tropics and Southeast Asia.

Impact of Initial Condition on ISO Prediction2

(a) skills of filtered rainfall initialized with the original NCEP_R2; (b) with doubled ISO signals in the NCEP_R2; (c) skills of filtered U850 initialized with the original NCEP_R2; (d) with doubled ISO signals in the NCEP_R2.

Signal CPL Forecast Error

ATM Forecast Error

Air-Sea Coupling Extends the Predictabilityof Monsoon Intraseasonal Oscillation

[ATM: 17 days; CPL: 24 days]Fu et al. (2007)

Monsoon Institute, Honolulu, January 07, 2008

* Waliser

et al. 2003

Signal: Mean amplitude of the ISO variance Forecast Error: Mean variance between ensembles

Control run Forecasts

L

Ljis X

Lij

20, )(

121

M

mij

mije XX

Mij1

20 )(1

Signal & Forecast Error

Potential predictability

Model Studies

NASA/GLA AGCM Waliser

et al. (2003)

ECHAM5 AGCM Liess

et al. (2005)

ECHAM4 AGCM, CGCM Fu et al. (2007)

Dynamical models has potential for ISO prediction

Predictability : Signal to Error Ratio Signal: Mean variance within ISO periodError: Mean variance between ensembles

~40 days

Signal

Error

* Liess

et al 2005

200h Pa Velocity Potential* Perfect model assumption

Potential predictability

0

5

10

15

20

25

30

35

40

45

50

C ER F EC M W IN G V LO D Y M AXP M ETF U KM O S N U 1 S N U 2 N C EP N AS A FS U 2 U H C AM

DEMETER APCC/CliPAS

Potential predictabilityNumber of forecast days until the signal equals the noise

From Kang

1

Better understand physical basis for intraseasonal

prediction.

Estimate 

potential 

and 

practical 

predictability 

of 

ISO 

in 

a 

multi‐model 

frame 

work. 

Developing 

optimal 

strategies 

for 

multi‐model 

ensemble 

(MME) 

ISO 

prediction 

system, 

including 

effective 

initialization 

schemes 

and 

quantification 

of 

the 

MME’s

ISO prediction skills with forecast metrics under operational conditions.

Identify 

model 

deficiencies 

and 

suggest 

ways 

to 

improve 

models’

convective 

and other physical parameterizations.

Revealing 

new 

physical 

mechanisms 

associated 

with 

ISV 

that 

cannot 

be 

obtained 

from analyses of a single model.

Study 

ISO’s 

modulation 

of 

extreme 

hydrological 

events 

and

its 

contribution 

to 

seasonal and interannual

climate variation. 

ISVHE Objectives

Experimental Designs

Free coupled runs with AOGCMs or AGCM simulation for at least 20 years

Daily or 6- hourly output

Control Run

ISO hindcast initiated every 10 for at least 45 days with more than 6 ensemble members from 1989 to 2008

Daily or 6-hourly output

ISV Hindcast EXP

Additional EXP for YoTC

period from May 2008 to Sep 2009

6-hourly output

YOTC EXP

1

CliPAS/ISVHE ParticipationsInstitution Participants

ABOM, Australia Harry Hendon, Oscar Alves

CMCC, Italy Antonio Navarra, Annalisa

Cherichi, Andrea Alessandri

CWB, Taiwan Mong-Ming Lu

ECMWF, EU Franco Molteni, Frederic Vitart

GFDL, USA Bill Stern

JMA, Japan Kiyotoshi

Takahashi

MRD/EC, Canada Gilbert Brunet, Hai

Lin

NASA/GMAO, USA S. Schubert

NCEP/CPC Arun

Kumar, Jae-Kyung E. Schemm

PNU, Korea Kyung-Hwan Seo

SNU, Korea In-Sik

Kang

UH/IPRC, USA Bin Wang, Xiouhua

Fu, June-Yi Lee

1

ECMWF

JMACWB

ABOM

ECNCEP

ISVHE MODELS Intraseasonal Variability Hindcast Experiment

The ISVHE is a coordinated multi-institutional ISV hindcast

experiment supported by APCC, NOAA CTB, CLIVAR/AAMP & MJO WG, NOAA CTB, and AMY.

UH IPRC

Supporters

SNUPNU

GFDL NASACMCC

http://iprc.soest.hawaii.edu/users/jylee/clipas.htm

1

BSISO

1.

Evaluation of models: (a) MJJAS variance distribution OLR  20‐90Days, (b) Interannual

variation of ISO variance, (3) 

Leading EEOF mode (PCC)

2.

Skill:  (a) Predictable mode forecast skill (PC1 and PC2), (b)  200hPa velocity potential, (c) precipitation and U850

3.

Predictability: Signal to noise ratio perfect model approach

4.

Prediction of interanual

variation statistical correction.

1 Evaluation on Control RunsPattern Correlation Coefficient and Normalized Root Mean Square Error for Mean Precipitation and 20-100-day Variance (30S-30N)

1 Evaluation on Control Runs20-100-Day U850, U200 and OLR along the equator (15oS-15oN)

The first two MV-EOF modes of 20-100-day 850-

and 200-hPa zonal

wind and OLR along the equator (15oS-15oN) obtained from obs

and

control simulations. The percentage variance explained by each mode is shown in the lower left of each panel. (Wheeler and Hendon 2004)

Preliminary results from 6 models

1 The MME and Individual Model Skills for MJO

Common Period: 1989-2008Initial Condition: 1st day of each month from Oct to MarchMME1: Simple composite with all modelsMMEB2: Simple composite using the best two modelsMMEB3: Simple composite using the best three models

1 MJO Skill depends on ENSO phase

Taking into account IAV anomaly, the practical TCC skill for the RMM1 and 2 extends about 5 to 10 days depends on model. The improvement is remarkable for the RMM1.

The skill in La Nina years is better than El Nino years in most models.

Any comments and questions?

Thank you !