Post on 16-Jan-2016
transcript
Tropical Axisymmetric Mode of Variability in the Atmospheric Circulation
UH seminar, 29 August 2001
Masahiro Watanabe Department of Meteorology, SOEST, University of Hawaii
* On leave from Center for Climate System Research (CCSR), University of Tokyo
refs: Watanabe, Kimoto, and Jin (2001, submitted to JC) Watanabe, Jin, and Kimoto (2001, to be submitted to JC)
introduction
• motivation of the study– What is the leading mode of the atmospheric circulation?
• e.g. barotropic & baroclinic instabilities, teleconnection patterns, MJO,….
– Attempt to specify and understand a principal mode in the global circulation fields ( teleconnection patterns in hemispheric fields)
• outline– observational data analysis
– AGCM simulation
– linear model diagnoses
UH seminar, 29 August 2001
previous studies
• near zonally uniform pattern (superrotational flow) – (e.g. Kang & Lau 1994)
• coherence with atmospheric angular momentum– (e.g. Anderson & Rosen 1983; Rosen & Salstein 1983)
• found in intraseasonal time scale – (e.g. Weickmann et al. 1997)
UH seminar, 29 August 2001
EOF1(23%) for monthly300, 1949-99
UH seminar, 29 August 2001
Tropical Axisymmetric Mode (TAM)= ‘global mode’ (Higgins et al. 2000; Bell & Halpert 2000) = ‘tropical mode’ (vonStorch 1999)
principal mode in
UH seminar, 29 August 2001
time series
PC1 M LOD Niño3 PC1 .83 .53 .47 M .58 .57LOD .41Niño3
(defined as the TAM index)
UH seminar, 29 August 2001
Regression of monthly NCEP anomalies on the 300 PC1
structure of TAM
UH seminar, 29 August 2001
ResidualTAM
R EOF1 explains19% of total variance,significantly correlatedwith M (0.62) andLOD (0.31)
TAM in the zonal-mean winds
UH seminar, 29 August 2001
spectral characteristics
NCEP
NCEP
AGCM
UH seminar, 29 August 2001
(=TAM index)
persistence of the TAM
UH seminar, 29 August 2001
question (1)
• The features of the TAM– high correlation with Niño3 SST index
– low-level divergence (convergence) over the maritime continent (eastern eq. Pacific)
– spectral peak around 4 yr
– persistence up to 5 mo
• Do they imply the TAM nothing more than the atmospheric response to El Niño?
• But ENSO residual fields do reveal the same variability dominating
• In essence, is TAM independent of ENSO?– How can we explore it? use AGCM !
UH seminar, 29 August 2001
T42L20 CCSR/NIES AGCM, 50yr run with climatological SST
TAM simulated by an AGCM
CCSR/NIES AGCM
UH seminar, 29 August 2001
simulated TAM
zonal wind
meridional wind
CCSR/NIES AGCM
UH seminar, 29 August 2001
question (2)
• AGCM reproduced an overall feature of the observed TAM– e.g. horizontal circulation patterns, even the equatorial surface wind
• spectrum of the coefficient is much whiter than observations in addition to the absence of a peak around 4yr
• TAM may essentially be an internal atmospheric mode– What is the dynamics responsible for such a mode?
– We need to diagnose it using a simpler dynamical framework
• Linearized multi-level PE model– derived from a dynamical core of the CCSR/NIES AGCM
– spectral truncation of T21
– vertical 20 levels
• steady version– zonally symmetric basic state obtained from the NCEP or AGCM cl
imatology (monthly or seasonal mean)
– zonal wavenumber truncated at m=5
• time integration– 3D basic state obtained from the NCEP or AGCM climatology (mo
nthly or seasonal mean)
– refs: Watanabe & Kimoto (1999, GRL)– Watanabe & Kimoto (2000, QJRMS)
UH seminar, 29 August 2001
linear baroclinic model (LBM)
fXXFLtX
fXXFXXLtX
XXFXXL
XXXXX
PTX
aaa
caaca
caac
ac
acac
s
feedbackeddy w/ flow Zonal3.
,
model response Zonal2.
,
model wavePlanetary 1.
anomaly : y,climatolog : wave,: mean, zonal :
ln,,,
**
****
*
**
zonal-wave coupling term
stationary wave feedback
UH seminar, 29 August 2001
linear operators
=
, D, T, ...
Xa
Xa*
m=0
m=1
m=2
m=M
, D, T,
ZRM
PWM
N(N~ 30,000 for T21L20)
F*(Xa,Xc*)
F(Xa*,Xc*)
UH seminar, 29 August 2001
linear operator matrices
calculate singular vectors of L
associated stationary wave anomalies readily obtained as
X L F X va c* * * *( , ) 1
i
(3)
(4)
1 2 3
1 2 3
L U V
X L f
U u u u
V v v v
u fa
v
T
, , ,
T
( , , ,...),( ...),( , , ,...),
( , )
1 2 3
1
ii
i
i
u-vectorssingular valuesv-vectors
∴ singular mode withthe smallest will have the longest persistence
UH seminar, 29 August 2001
neutral mode detection
UH seminar, 29 August 2001
neutral mode
zonal wind
meridional wind
UH seminar, 29 August 2001
Leading singular mode + associated stationary waves, v1+L*-1F*(Xc*,v1)
・ much prevailing zonal structure in 300
・ low-level features less similar to obs./AGCM TAM・ decay time ~ dissipation timescale of the free troposphere
(< month)
anomalous circulation associated with the neutral mode
UH seminar, 29 August 2001
zonal asymmetry
observed TAM
neutral mode
Ua
・ neutral mode seems consistent with the observed TAM in a considerable part except for the Pacific
UH seminar, 29 August 2001
Zonal-mean zonal momentum budget
close to neutrality
on the neutrality of the mode
UH seminar, 29 August 2001
NCEP zonal-mean wind regressed on the PC1 300
Coincidence between Ua and c further suggests themomentum feedback actively working for the neutrality
role of the basic state vorticity
UH seminar, 29 August 2001
eigenmodes of the zonal-mean shallow-water eqs.
・ basic state is not crucial for the presence of the mode・ scattering on i=0, due to viscosity?
origin of the neutral mode
conclusions (1)
• Tropical Axisymmetric Mode (TAM):– tightly related to the angular momentum variability and LOD
– contains a signature of El Niño (may suggest ENSO forces TAM)
• dynamics of the TAM– AGCM with climatological SST does reproduce the observed TAM
– A near-neutral mode found in the singular mode computation of the linear model is considerably similar to the observed/AGCM TAM
– The essence of the TAM can be interpreted as an internal atmospheric mode which is easily excited by forcing
– The neutrality partially arises from a positive momentum feedback in the zonal mean state (i.e. coupling between Ua and Hadley circulation) , although the process may not be crucial for the origin of mode
UH seminar, 29 August 2001
question(3)
UH seminar, 29 August 2001
• Neutral mode failed to reproduce the lower-tropospheric feature in the observed TAM– Why?
– Interaction between convection and the dynamics?
UH seminar, 29 August 2001
Composite OLR anomaly based on the TAM index
AGCM
NOAA
convection associated with TAM
implication
UH seminar, 29 August 2001
• Can we interpret the change in zonal mean state during ENSO in terms of an excitation of the neutral mode?
• Role of the zonal mean flow (Ua) in:– ENSO upstream teleconnection
– ENSO-monsoon coupling
UH seminar, 29 August 2001
Regression of Z500/300 on monthly Nino3 SSTA, 1949-99
global ENSO teleconnection
UH seminar, 29 August 2001
DJFidealized heating Q
ENSO-forced zonal-mean flow
The Ua response is independent of the Rossby wave train over the Pacific!
NCEP composite LBM response
calculate singular vectors of L (zonal-mean dynamical operator)
The phase and amplitude of each mode depend not only on the singular value, but on the projection of u-vector onto forcing
(3)
(4)
1 2 3
1 2 3
L U V
X L f
U u u u
V v v v
u fa
v
T
, , ,
T
( , , ,...),( ...),( , , ,...),
( , )
1 2 3
1
ii
i
i
u-vectorssingular valuesv-vectors
UH seminar, 29 August 2001
zonal mean response represented by singular modes
‘projection coefficient’
UH seminar, 29 August 2001
・ a large part of the forced zonal wind is reproducible with two singular modes・ different optimal heating profiles for the neutral mode (~ TAM) & a second (baroclinic) mode
optimal thermal forcing
reconstruction by singular modes
( ) (1) ( ) (2)
L X X F X X fL X X F X X f
c a c a
c a c a
( , )( , )
* *
* * * * *
zonal-wave coupling
El Niño heating
UH seminar, 29 August 2001
stationary wave response to ENSO-forced Ua
T850 in winterNCEP composite for El Niño
LBM response to Q
95% significance
idealized heating
LBM response to Q+ZW
UH seminar, 29 August 2001
‘tropical-belt’ teleconnection
upslope cooling (downslope warming) due to orographic forcing (cf. Hoskins & Karoly 1981)
UH seminar, 29 August 2001
time series of : TAM index (JJA avg.), all-India monsoon rainfall (IMR),
Webster & Yang ‘s dynamical monsoon index
r(TAM,IMR) = -0.50r(TAM,DMI) = -0.62
Relationship between TAM and summer monsoon
question (4)
UH seminar, 29 August 2001
• There is an argument that change in the subtropical jet associated with El Niño is involved in the coupled ENSO-monsoon system.– (e.g. Nigam 1994; Ju & Slingo 1995)
• The TAM index indeed shows a significant correlation with indices of the Asian summer monsoon variability
• Does the anomalous zonal-mean flow forced by El Niños (whatever the mechanism) plays any role in the ENSO-monsoon coupling?
UH seminar, 29 August 2001
Composite OLR anomaly in summer, following Kawamura (1998)
convection associated with ENSO/monsoon
weak monsoon/warm event: 1979, 1983, 1987, 1991, 1992, 1993strong monsoon/cold event : 1981, 1984, 1985, 1988, 1989, 1990
UH seminar, 29 August 2001
precursors for weak summer monsoon
observed composite in May
OLR anomaly
T300 & V850 anomalies
UH seminar, 29 August 2001
simulated circulation anomalies in May
10-member ensemble difference for El Niño run
Vertically averaged Q
T300 & V850 response
CCSR/NIES AGCM
UH seminar, 29 August 2001
simulated monsoon precursor in May
CCSR/NIES AGCM
UH seminar, 29 August 2001
role of Ua in forcing the continental cooling
T300 & V850 response
to the AGCM heating
LBM at day 25
Indian Ocean heating removed
Indian Ocean heating removed& zonal mean response damped
UH seminar, 29 August 2001
NCEP composite
cooling over the Himalayan upslope
CCSR AGCM
LBM (day 25)
temperature longitude-pressure section along 30N in May
vertical phase tilt
upslope cooling forlong wave (K<Ks)
UH seminar, 29 August 2001
cooling over the Himalayan upslope
Cp1
p0
C
<0
conclusions (2)
• zonal flow anomaly (Ua) during ENSO– subtropical westerlies and equatorial easterly anomaly
– the anomalous zonal-mean flow is plausibly independent of the Rossby wave train over the Pacific
– may be an indication of neutral modes excited by the El Niño heating
• role of zonal flow anomaly: tropical-belt teleconnection– Ua-induced teleconnection seems to explain how and why the anomalo
us circulation occurs in the upstream region of El Niño
– tropical-belt teleconnection may further play an active role in the ENSO-monsoon coupling such that it helps to precondition the weak monsoon during El Niño-like condition
• further question: how ENSO forces the zonal mean flow anomaly?– what is the role of stationary wave feedback on to the zonal mean?
UH seminar, 29 August 2001