Page 1
Interannual Variability of Indian Summer Monsoon arising from Interactionsbetween Seasonal Mean and Intraseasonal Oscillations
E. SUHAS, J. M. NEENA, AND B. N. GOSWAMI
Indian Institute of Tropical Meteorology, Pashan, Pune, India
(Manuscript received 29 July 2011, in final form 29 November 2011)
ABSTRACT
A significant fraction of interannual variability (IAV) of the Indian summer monsoon (ISM) is known to be
governed by ‘‘internal’’ dynamics arising from interactions between high-frequency fluctuations and the
annual cycle. While several studies indicate that monsoon intraseasonal oscillations (MISOs) are at the heart
of such internal IAV of the monsoon, the exact mechanism through which MISOs influence the seasonal mean
monsoon IAV has remained elusive so far. Here it is proposed that exchange of kinetic energy (KE) between
the seasonal mean and MISOs provides a conceptual framework for understanding the role of intraseasonal
oscillations (ISOs) in causing IAV and interdecadal variability (IDV) of the ISM. The rate of KE exchange
between seasonal mean and ISOs is calculated in frequency domain for each Northern Hemispheric summer
season over the ISM domain, using 44 yr of the 40-yr ECMWF Re-Analysis (ERA-40) data. The seasonal
mean KE and the rate of KE exchange between seasonal mean and ISO shows a significant relationship at
both the 850- and 200-hPa pressure levels. Since the rate of KE exchange between seasonal mean and ISO is
found to be independent of known external forcing, the variability in seasonal mean KE arising from this
exchange process can be considered as an internal component explaining about 20% of IAV and about 50%
of IDV. Contrary to the many modeling studies attributing the weakening of tropical circulation to the sta-
bilization of the atmosphere by global warming, this paper provides an alternative view that internal dynamics
arising from scale interactions might be playing a significant role in determining the decreasing strength of the
monsoon circulation.
1. Introduction
The dependence of agriculture, drinking water, and
energy production on the Indian summer monsoon (ISM)
rainfall makes it the lifeline for a large fraction of the
world’s population. The economy, life, and property in
the region are vulnerable to significant variability of the
ISM on intraseasonal, interannual, and interdecadal time
scales (Webster et al. 1998; Krishnamurthy and Goswami
2000; Goswami et al. 2006b). Hence, predicting the sea-
sonal mean ISM rainfall is of great socioeconomic im-
portance and has been attempted for many decades,
albeit with limited success (Gadgil et al. 2005; Kang and
Shukla 2006). Recognizing the fact that the tropical
climate is largely determined by changes in slowly varying
boundary forcing, a physical basis for prediction of sea-
sonal mean monsoon rainfall was proposed by Charney
and Shukla (1981). This concept, supported by a large
number of modeling studies (e.g., Shukla 1998), estab-
lished the basis for seasonal prediction in the tropics.
Such slowly varying forcing can be considered ‘‘external’’
for the atmosphere, and may be predictable as it arises
from the slow oscillations of the coupled climate system.
Some of the major external forcings—the El Nino–
Southern Oscillation (ENSO), Indian Ocean dipole
(IOD), Atlantic multidecadal oscillation (AMO), and
Eurasian snow cover—are known to influence the in-
terannual variability (IAV) and interdecadal variabil-
ity (IDV) of the Indian monsoon and have been well
documented. Initially identified from correlations, the
physical mechanisms and different pathways through
which these phenomena influence the Indian monsoon
have been further elucidated through diagnostic stud-
ies and model simulations (Rasmusson and Carpenter
1983; Shukla and Paolino 1983; Webster et al. 1998;
Krishnamurthy and Shukla 2008; Goswami et al. 2006a;
Krishnamurthy and Goswami 2000; Krishna Kumar
et al. 2006; Hahn and Shukla 1976).
Corresponding author address: E. Suhas, Indian Institute of
Tropical Meteorology, Dr. Homi Bhaba Road, Pashan, Pune
411008, India.
E-mail: [email protected]
VOLUME 69 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S JUNE 2012
DOI: 10.1175/JAS-D-11-0211.1
� 2012 American Meteorological Society 1761
Page 2
While some fraction of the IAV of the seasonal mean
may be governed by such predictable external forcing,
a fraction of the IAV of the seasonal mean may arise
from interactions among different scales of motion, in-
teraction between seasonal mean and ISOs, interaction
between organized convection and large-scale circula-
tion, interaction between flow and topography, and so
on. The contribution to the IAV of seasonal mean
arising from these processes is collectively called the
‘‘internal’’ component. It has been argued that the IAV
of seasonal mean caused by the internal component is
less predictable (Goswami 1998). Hence, the potential
predictability of the seasonal mean would depend on the
relative contribution of the external and internal com-
ponents to the IAV. Recent studies using both obser-
vational analysis and model simulations (Ajaya Mohan
and Goswami 2003; Kang et al. 2004; Goswami and
Xavier 2005; Krishna Kumar et al.2005) show that the
Asian monsoon region is an exception in the tropics
where the potential predictability is relatively low, un-
derpinning the role of internal variability compared to
external variability over this region. The seminal role of
internal variability in determining the seasonal mean
may be one reason for the current poor skill of almost all
models in predicting the seasonal mean ISM rainfall
(Kang and Shukla 2006). A clear understanding of the
mechanism responsible for the internal IAV over the
ISM domain may help us develop better initialization
techniques for improving the seasonal forecast.
The intraseasonal oscillations are quasi-periodic fluc-
tuations of atmospheric origin, which play a major role
in determining the amplitude of seasonal mean of in-
dividual summer seasons by modulating the strength
and duration of active/break spells of the ISM through
the northward-propagating 30–60-day mode and the
westward-propagating 10–20-day mode (Goswami 2005,
and references therein). The ISOs are also found to exert
an indirect control on the seasonal mean by modulating
the synoptic-scale activity through clustering of lows and
depressions along the monsoon trough (Goswami et al.
2003). Hence, the IAV of monsoons arising from in-
teraction between seasonal mean and ISOs may be
a major source for internal IAV. The internal IAV has
received much attention recently and a few studies have
proposed mechanisms for the generation of internal IAV
involving summer ISOs (Sperber et al. 2000; Goswami
and Ajaya Mohan 2001; Goswami et al. 2006b; Hoyos and
Webster 2007). Since the dominant modes of IAV of the
seasonal mean and the ISOs share a common spatial
pattern, it is hypothesized that a shift in the probability
density function of ISOs can give rise to some internal
IAV (Ferranti et al. 1997; Sperber et al. 2000; Goswami
and Ajaya Mohan 2001). It has also been argued that
nonlinear interaction between the ISOs and the annual
cycle could give rise to a biennial internal variability of
the monsoon (Goswami 1995; Goswami et al. 2006b).
While these studies demonstrated the proof of the con-
cept using simple heuristic models, the quantitative con-
tribution of such interactions to the observed IAV could
not be estimated. Several studies showed the existence
of significant correlation between the seasonal mean
and the ISO over the ISM domain, indicating a physical
relationship between these two scales (Lawrence and
Webster 2001; Qi et al. 2008; Fujinami et al. 2011).
Corroborative evidence in this direction is also obtained
by estimating the correlation between area-averaged
variance of ISO and seasonal mean rainfall over India for
104 yr (correlation coefficient 5 0.53). The correlation is
much higher during the recent 54-yr period (correlation
coefficient 5 0.64) (Fig. 1a). Although all these studies
provide circumstantial evidence for the link between
ISOs and seasonal mean over the ISM domain, no clear
picture of how ISOs lead to internal IAV has emerged so
far. Thus, a physical mechanism through which the ISOs
influences the seasonal mean and introduces internal
IAV still remains an open question.
Recently, Qi et al. (2008) made an attempt to quantify
the role of ISO perturbations in determining the strength
of the mean zonal wind, and thereby the strength of ISM,
through a dynamical framework. The relationship be-
tween the mean zonal wind tendency and the eddy
momentum transport was derived from the primitive
equations of motion. It is shown that the eddy momentum
transport can significantly affect the westerly tendency
during strong and weak ISO years. However, since the
eddy momentum transport term includes variability on
all time scales, the role of ISO feedback in affecting the
strength of the mean westerly is not clear. Moreover, the
seasonal mean variability of ISM rainfall is more strongly
related to the strength of the winds over the low-level
jet (LLJ) and the upper-level tropical easterly jet (TEJ)
regions (Figs. 1b,c). Besides, not only the zonal wind
but also the meridional wind component is significant.
Figure 2a shows the seasonal mean rainfall averaged over
the Indian landmass (6.58–27.58N, 72.58–85.58E) and the
seasonal mean kinetic energy (KE) at 850 hPa averaged
over the LLJ core region (08–208N, 508–708E). A strong
correlation of 0.61 significant at the 95% confidence level
establishes the connection between the seasonal mean
rainfall and the cross-equatorial flow. Figure 2b shows the
seasonal mean rainfall averaged over the landmass and
the seasonal mean KE at 200 hPa averaged over the TEJ
region (08–158N, 508–808E). The moderately strong cor-
relation coefficient (0.52, significant at the 95% confidence
level) suggests a significant relationship between the
strength of upper-level monsoon circulation and seasonal
1762 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69
Page 3
mean monsoon rainfall. The strong relationship between
the IAV of KE of LLJ and that of TEJ and monsoon
rainfall implies that better insight into the IAV of mon-
soon rainfall can be obtained through understanding
the mechanisms of IAV of LLJ KE and TEJ KE. To
quantify the role of interaction between seasonal mean
and ISOs in causing the IAV and IDV of the seasonal
mean, the analysis scheme presented by Qi et al. (2008)
is reframed using spectral energetic analysis. In the
present study, we investigate how scale interactions be-
tween the seasonal mean monsoon and the ISOs can give
rise to variability of kinetic energy of the monsoon flow
on interannual and interdecadal time scales.
The data used in this study are described in section 2.
A dynamical framework for the ISO–seasonal mean
interaction in terms of atmospheric energetics is given in
section 3. Details of the computation of KE exchange
between the seasonal mean and ISO is presented in
section 4. Section 5 describes the IAV and IDV of the
exchange of KE between the seasonal mean and ISOs,
and its relationship with the IAV and IDV of seasonal
mean. The issue of sensitivity of temporal window length
on the calculation of seasonal mean–ISO interaction is
addressed in section 6. The dependence of the KE ex-
change on the reanalysis data used is discussed in section 7.
Section 8 summarizes the important results.
2. Data
Our primary analysis is based on 44 yr (1958–2001) of
the 40-yr European Centre for Medium-Range Weather
Forecasts (ECMWF) Re-Analysis (ERA-40) daily wind
data (Uppala et al. 2005) at the 850- and 200-hPa pres-
sure levels for the ISM period [i.e., June–September
(JJAS)]. The consistency of the analysis is tested with
National Centers for Environmental Prediction (NCEP)–
National Center for Atmospheric Research (NCAR)
reanalysis daily wind data (Kalnay et al. 1996). We have
also used high-resolution 18 3 18 daily rainfall data
available for the period 1901–2004 (Rajeevan et al. 2008).
Monthly Nino-3.4 data available from 1871 to 2007 are
downloaded from the NCAR Earth System Laboratory
(http://www.cgd.ucar.edu/cas/catalog/climind/TNI_N34/)
and the JJAS mean was calculated. The Pacific de-
cadal oscillation (PDO) index is downloaded (from
FIG. 1. (a) Seasonal mean rainfall (mm day21, solid line) and
JJAS ISO (10–90-day bandpass filtered rainfall, dashed line) vari-
ance averaged over the region 10.58–27.58N, 72.58–85.58E from
1901 to 2004 (data source: India Meteorological Department
(IMD) rainfall data; Rajeevan et al. 2008). (b),(c) Climatological
seasonal mean (JJAS) kinetic energy per unit mass (m2 s22) cal-
culated for the period 1958–2001 using the ERA-40 data (Uppala
et al. 2005) at (b) 200 and (c) 850 hPa. The domain of analysis used
in the study at both levels is highlighted with a rectangular box.
FIG. 2. Seasonal mean (JJAS) rainfall (mm day21, solid line) and
seasonal mean kinetic energy per unit mass (m2 s22) calculated
using ERA-40 data (dashed line) (a) at 850 hPa averaged over
08–208N, 508–708E and (b) at 200 hPa averaged over 08–158N,
508–808E. The corresponding correlation coefficients (CC) are
given at the upper-left corner of each panel.
JUNE 2012 S U H A S E T A L . 1763
Page 4
http://www.atmos.washington.edu/;mantua/; Mantua
et al. 1997), as is the dipole mode index (DMI) data
(representative of IOD; see http://www.jamstec.go.jp/
frsgc/research/d1/iod/). A North Atlantic SST index is
calculated by area averaging the anomalies of Kaplan
SST V2 data, provided by the National Oceanic and
Atmospheric Administration Office of Oceanic and
Atmospheric Research, Earth System Research Labo-
ratory, Physical Science Division (NOAA/OAR/ESRL
PSD) at Boulder, Colorado (see http://www.esrl.noaa.
gov/psd/) over the region 08–708N, 908W–208E. An AMO
index is calculated by applying 11-yr running mean on the
JJAS mean of the North Atlantic SST index (Goswami
et al. 2006a).
3. Dynamical framework for understanding theISO–seasonal mean interaction
The dynamic effect of transient eddies on the time
mean flow received considerable attention during the
1970s and 1980s (Holopainen 1978a; Hoskins et al. 1983;
Hoskins and Pearce 1983). Most of these studies focused
on the influence of transient disturbances on the local
time mean flow over the midlatitude regions in terms of
large-scale Reynolds stresses in a quasigeostrophic (QG)
framework. The atmospheric flow at any time period can
be partitioned as the sum of time mean and transient
fluctuations, and this concept was used by Holopainen
(1978b) to derive the growth or decay of time mean KE
using the primitive equations of motion. The rate of
change of time mean KE is the resultant of various
physical processes. The abbreviated representation of
the equation for the rate of change of KE is given below
(the complete forms of the equations are given in the
appendix):
d(K)/dt 5 BC 1 CAK 1 WF 1 D, (1)
where K represents the time mean KE, BC represents
the barotropic energy conversion, CAK stands for con-
version of available potential energy (APE) to KE, and
WF and D indicate the wave energy flux and dissipation
due to friction, respectively. The term CAK acts as the
source for KE through conversion from APE generated
by the diabatic heat sources in the tropics. The wave
energy flux term includes the advection of KE as well as
the work done by pressure. The BC term involves eddy–
time mean interaction. The mechanism for the growth or
decay of KE in the time mean scale can be summarized
as follows.
The net radiation balance and the resultant equator-
to-pole temperature gradient maintain a thermally
direct large-scale circulation (Hadley circulation) in the
tropics (Held and Hou 1980; Schneider and Lindzen
1977). Such a circulation in the presence of abundant
moisture favors large-scale deep convection in the tropics.
Deep convective activity generates APE that gets con-
verted to KE in various scales. The time mean may gain
KE through this direct conversion. Besides, the KE of
the time mean scale over a local domain may result from
barotropic energy conversion involving eddy–time mean
interactions and from advection of KE by time mean and
eddies. This is the basic principle behind the analysis
used in this study.
This framework has been widely used by many studies
to quantify the effect of transient eddies over differ-
ent tropical regions (Lau and Lau 1992; Maloney and
Hartmann 2001; Seiki and Takayabu 2007; Serra et al.
2008). The growth of barotropic eddy KE through eddy–
mean flow interaction over the eastern and western Pa-
cific was studied by Maloney and Hartmann (2001), and
this study noted the role of the Madden–Julian oscillation
(MJO) in making favorable conditions for the formation
of tropical cyclones over the Pacific. The eddy momen-
tum transport term used in the study by Qi et al. (2008)
can be viewed as an approximate form of the term BC.
The interaction between transient eddies and time mean
depends on the averaging period. The transient fluctua-
tions include a collection of time scales whose maxima
and minima depend on the time resolution and length of
the averaging period. The goal of the present study is to
quantify the feedback between the ISO and seasonal
mean time scale over the ISM domain. Such a feedback
involves interactions between the time mean and multi-
ple frequencies in the ISO scale. Hence in the place of
eddy–time mean interaction framework, we study the
scale interactions in the spectral framework.
The spectral energetic analysis method was originally
introduced by Saltzman (1957) in the wavenumber do-
main for estimating the eddy KE contribution coming
from different spatial scales. This formulation led to
several case studies examining the role of KE ex-
change between zonal mean and waves, mostly the role
of planetary-scale waves in the extreme behavior of the
monsoon (Kanamitsu et al. 1972; Awade et al. 1982).
However, the monsoon ISO is defined as the fluctuations
whose time scale is greater than synoptic scale but less
than the seasonal cycle; the energetic analysis in the
wavenumber domain is not suitable for extracting the
information about interactions between time mean and
monsoon ISO over a region. Hayashi (1980) modified
Saltzman’s scheme using a cross-spectral technique, and
it can be used for studying the interactions between
different scales of motion and the spectral transfer of
energy in both wavenumber and frequency space.
1764 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69
Page 5
Hayashi’s analysis scheme decomposes the KE equation
in the frequency domain, and the energy conversions
among different time scales can be studied through
spectral windows of different frequency ranges. A com-
prehensive study of the spectral energetics of the atmo-
spheric circulation in the frequency domain was carried
out by Sheng and Hayashi (1990). Since then, this method
has been extensively used for quantifying the interactions
among different time scales ranging from synoptic to in-
terannual over different tropical domains (Krishnamurti
et al. 2000, 2003; Krishnamurti and Chakraborty 2005;
Neena and Goswami 2010). In this study, we adopt an
energetic analysis in the frequency domain to gain insight
into the mechanism of interaction between ISO and
seasonal mean of the ISM and the IAV caused by this
process.
4. Kinetic energy exchange between seasonalmean and ISO
The growth or decay of KE in the frequency domain
can be derived by applying Fourier analysis in time on
the zonal and meridional momentum equations in
spherical coordinate and by solving the resultant equa-
tions with the continuity equation. Analogous to Eq. (1)
in the time domain, the time rate of change of KE in the
frequency domain can be represented by Eq. (2):
›Kn
›t5 h(Km, Kp) �Kni 1 hK �Kni 1 hAn �Kni 1 Fn.
(2)
The rate of KE of a given scale n Kn involves four
different physical processes. The first process represents
the transfer of KE to the scale of frequency n from
a couple of other frequencies m and p that are governed
by a trigonometric selection rule, namely n 5 m 1 p or
n 5 jm 2 pj for an exchange. The second process is the
growth (or decay) of KE of the given frequency n when
it interacts with K. The third process is the growth (or
decay) of KE of a given frequency from the eddy
available potential energy An at the same scale. The last
term Fn represents the net loss or gain of KE of fre-
quency n by friction.
We adopt the method proposed by Hayashi (1980) for
calculating the exchange of KE between the seasonal
mean and individual frequencies. A positive (negative)
sign for (K, Kn) implies that the seasonal mean loses
(gains) energy to (from) the frequency n. The terms
involved in the exchange calculation at a particular
pressure level are given as follows:
hK �Kni 5 2
�›u
›xPn(u, u) 1
›y
›xPn(u, y)
�|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
A
2
�›u
›yPn(u, y) 1
›y
›yPn(y, y)
�|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
B
2tanu
a[uPn(u, y) 2 yPn(u, u)]|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
C
2
�›(uKn)
›x1
›(yKn)
›y1
›(vKn)
›p
�|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
D
2
�›u
›pPn(u, v) 1
›y
›pPn(y, v)
�|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
E
, (3)
where u, y, and v respectively represent the zonal, me-
ridional, and vertical wind components in pressure co-
ordinates and u, y, and v are their seasonal mean
counterparts. Also, Pn and Kn are the cross-spectral co-
efficient of frequency n and KE of nth frequency, re-
spectively; u is the latitude and a is the radius of the earth.
It can be inferred from Eq. (3) that the calculation of
the rate of KE exchange between seasonal mean and
any scale n involves five components (A–E). First, two
terms involves the horizontal gradient of seasonal mean
u wind, seasonal mean y wind, and the cross-spectral
coefficients. The third term represents the product of the
curvature term, seasonal mean u and y winds, and the
cross-spectral coefficients. The fourth term is the con-
vergence of KE of scale n due to seasonal mean wind.
The last term (E) implies the baroclinic processes through
which the scale n and the time mean interact. Since the
order of magnitude of the terms involving the vertical
derivative is very small, it is neglected. This formulation
can be used for examining the rate of KE exchanges
among different scales over a local domain, which would
be helpful in understanding the fundamental dynamics of
interactions between different scales over the region.
The rate of KE exchange between seasonal mean and
each frequency during the 1958–2001 summer seasons is
calculated using JJAS daily wind data of length 122 days.
With 122-day data for each season, 61 frequencies are
resolved. Grouping those frequencies that come under
the ISO scale (10–60-day periodicity, 2–12 harmonics),
the sum of KE exchange by all of these frequencies
with the seasonal mean is considered as the net KE ex-
change by the ISOs with the seasonal mean. The har-
monics representing 122-day periodicity is not considered
as part of ISO since it is close to the seasonal cycle. The
JUNE 2012 S U H A S E T A L . 1765
Page 6
rate of KE exchange between the seasonal mean and
ISO for 44 summer seasons is calculated using ERA-40
wind data at 200 hPa over the region 08–158N, 508–
808E, where a significant relationship exists between the
wind strength and the seasonal mean Indian monsoon
rainfall. Figure 3a shows the IAV of rate of KE ex-
change between the seasonal mean and ISO (thick black
line) at 200 hPa and the contributions of different terms
in Eq. (3) to this exchange. While during most years the
ISOs take energy from the seasonal mean, the net ex-
change also exhibits considerable IAV. The conver-
gence of ISO KE due to seasonal mean zonal wind
dominates the net rate of KE exchange at 200 hPa. It
explains about 43% of the total variability. Similar to the
net rate of KE exchange, the convergence of ISO KE
due to seasonal mean u wind also gives energy to the
ISO scale. No notable trend is observed in this exchange.
The convergence of ISO KE due to meridional wind is
always negative, which means that energy is transferred
to the seasonal mean scale. A decreasing trend is ob-
served in this exchange, which implies that the seasonal
mean is losing more energy to the ISO in recent decades.
The sum of the first, second, third, and 2d(V0Kn)/dy
terms cancel out each other, making the effective con-
tribution of these terms nearly zero and leaving the last
term [2d(U0Kn)/dy] to dominate the net KE exchange.
The same analysis is repeated at 850 hPa over the LLJ
core region that shows a strong relationship with the
Indian monsoon rainfall. Figure 3b shows the net rate of
KE exchange between seasonal mean (thick black line)
and ISO and the contributions from different terms in
Eq. (3) to this exchange. One interesting fact to be noted
in Figs. 3a and 3b is the opposite sign of the 2d(U0Kn)/dx
and 2d(V0Kn)/dy components in the upper and lower
atmosphere. It may be related to the opposing directions
of circulation in the lower and upper levels. If the rate of
KE exchange between seasonal mean and ISO were
significant to make changes in the amplitude of seasonal
mean KE, it would be reflected in the variability of
seasonal mean KE and there would be a significant
phase relationship between seasonal mean KE and the
rate of KE exchange between seasonal mean and ISO.
In the next section, evidences for significant internal
interannual and interdecadal variability are presented.
5. Internal interannual and interdecadal variabilityof the ISM
To quantify the variability of the ISM IAV associated
with the central Pacific SST anomaly, seasonal mean KE
at 200 hPa over the TEJ region is correlated with the
seasonal mean Nino-3.4 index for the 1958–2001 period.
The Nino-3.4 index represents the area-averaged SST
anomaly over the region 58S–58N, 1208–1708W. The cor-
relation coefficient between the two is 20.53, significant
at the 95% confidence level. Thus, the ENSO can explain
about 30% of the IAV of seasonal mean ISM (as repre-
sented by the KE of TEJ).
Although the ENSO is a dominant driver of IAV of
most of the climate systems around the globe, about
70% of the IAV of the ISM arises from processes other
than the ENSO. External forcing such as IOD and the
North Atlantic SST anomaly could together explain
another 30% of the ISM IAV (Table 1). Hence, it is
suspected that the internal processes can account for
some of the unexplained variability. The rate of KE ex-
change between the seasonal mean and ISO (10–60 day)
at 200 hPa is calculated over the same region for the same
time period using ERA-40 data. Figure 4a shows the IAV
of seasonal mean KE at 200 hPa and the rate of KE ex-
change between seasonal mean and ISO over the TEJ
region. It can be inferred from the figure that almost all
of the years during which the seasonal mean KE shows
a local maximum, ISO supplies energy to the seasonal
FIG. 3. Different components of rate of kinetic energy exchange
per unit mass between the seasonal mean and 10–60-day time scale
(31025 W kg21) (a) at 200 hPa over the region 08–158N, 508–808E
and (b) at 850 hPa over the region 08–208N, 508–708E. The solid
black line represents the net exchange, the dotted line with squares
represents the sum of first three terms of Eq. (3), the dashed line
with open circles represents the exchange due to the 2d(U0Kn)/dx
term, and the dot-dashed line with solid circles represents the ex-
change due to the 2d(V0Kn)/dy term.
1766 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69
Page 7
mean. At the same time, whenever the seasonal mean
KE is low, either the ISO takes energy away from the
seasonal mean or contributes little to the seasonal mean.
Thus, it clearly demonstrates that internal processes
(i.e., energy exchange among scales) can significantly
affect the amplitude of seasonal mean and thereby cause
its variability on interannual time scale (Hoyos and
Webster 2007). The correlation coefficient between the
KE of seasonal mean and the KE exchange between ISO
and seasonal mean is 20.42, significant at the 95% con-
fidence level. Despite exhibiting a significant relationship
with the seasonal mean KE, the ENSO and other sources
of external forcing do not exhibit any relationship with
the energy exchange processes, indicating that the pro-
cess is purely internal (Table 1). The ENSO effect is
removed from the seasonal mean KE by regressing
it with the Nino-3.4 index. The ENSO removed sea-
sonal mean KE shows enhanced relationship (correla-
tion coefficient 5 20.48) with the energy exchange. Hence,
it may be concluded that the internal process can explain
about 20% of the IAV of seasonal mean of the ISM.
Since the South Asian monsoon is a result of baroclinic
response of the atmosphere to deep tropospheric con-
vective heating, the lower-level circulation is expected to
be related to the upper-level circulation. Hence, a similar
relationship can be expected to exist in the lower atmo-
sphere between the KE of seasonal mean and its ex-
change with the ISO. The seasonal mean KE at 850 hPa
over the LLJ core region and the rate of KE exchange
between the seasonal mean and ISO are calculated for
the 1958–2001 period using ERA-40 data. Figure 4b
corroborates the finding at 200 hPa. It shows a corre-
lation coefficient of 20.28. Even though it is not as
strong as that in the upper atmosphere, the relationship
improves (correlation coefficient 5 20.31) when the
ENSO effect is removed from the seasonal mean KE by
linearly regressing it with the Nino-3.4 index (Table 1).
TABLE 1. Correlations at 850 and 200 hPa on interannual and interdecadal time scales. For the interannual time scale a correlation
coefficient greater than 0.25 is significant at 95% confidence level for 42 degrees of freedom. For the interdecadal time scale the signif-
icance is also estimated at 95% confidence level. The significant correlations are in boldface.
Correlation coefficient
Interannual time scale Interdecadal time scale
850 hPa 200 hPa 850 hPa 200 hPa
Seasonal mean KE and seasonal mean rainfall 0.61 0.52 — —
Seasonal mean KE and exchange 20.28 20.425 20.65 20.76
Seasonal mean KE and Nino-3.4 20.33 20.53 0.32 0.49
Exchange and Nino-3.4 20.095 0.03 0.04 20.6
ENSO removed seasonal mean KE and exchange 20.31 20.48 20.65 20.71Partial correlation coefficient between seasonal mean KE
and exchange excluding the influence of Nino-3.4
20.33 20.483 20.70 20.67
Seasonal mean KE and IOD 20.0574 20.39 0.43 0.169
Exchange and IOD 0.0405 20.0623 20.31 20.34
Seasonal mean KE and North Atlantic SST 0.127 0.39 — —
Exchange and North Atlantic SST 20.114 20.107 — —
Seasonal mean KE and AMO — — 0.41 0.79AMO removed seasonal mean KE and exchange — — 20.81 20.734
Seasonal mean KE and PDO — — 0.46 0.36
Exchange and PDO — — 20.47 20.4
Partial correlation coefficient between seasonal mean KE
and exchange excluding the influence of PDO
— — 20.55 20.72
FIG. 4. Seasonal mean kinetic energy per unit mass (m2 s22, solid
line) and the rate of kinetic energy (per unit mass) exchange be-
tween the seasonal mean and 10–60-day time scale (31025 W kg21,
dashed line) calculated using ERA-40 data (a) at 200 hPa over
the region 08–158N, 508E–808E and (b) at 850 hPa over the region
08–208N, 508–708E. The corresponding correlation coefficients (CC)
are given at the upper-left corner of each panel.
JUNE 2012 S U H A S E T A L . 1767
Page 8
The moderate relationship between seasonal mean KE
and the energy exchange process can be understood by
considering the fact that the 850-hPa atmosphere is close
enough to come under the influence of the planetary
boundary layer and scale separation in the noisy atmo-
sphere is difficult. In addition to that, the friction term
and boundary fluxes are neglected while calculating the
exchange over the ISM domain both in the upper and
lower atmosphere. The boundary flux terms become
zero when the equation is integrated over the globe, but
it is nonzero if the calculation is restricted to a small
area. The relationship between the seasonal mean KE at
850 hPa and the ENSO is also weak (it shows only
a moderate correlation coefficient of 20.33). Similar to
the upper atmosphere, the exchange process shows no
relationship with the ENSO. The exchange process and
its phase relationship with the seasonal mean KE at
850 hPa validate our finding that the scale interaction is
purely internal, and it can explain considerable amount of
IAV of the ISM.
Since the KE exchange between ISO and seasonal
mean is strongly related to the seasonal mean KE, any
long-term tendency of this exchange may lead to longer-
term variability of the ISM. The internal IDV of the ISM
can be brought out by applying an 11-yr running mean
on both the seasonal mean KE and the exchange. The
running mean is the simplest available low-pass filter
that removes fluctuation whose periodicity is less than its
window size by preserving the low-frequency oscillation.
However, when computing the correlation between two
time series subject to low-frequency filtering, one has to
be cautious about the reduced number of degrees of
freedom. Filtering has the effect of smoothing the time
series and thereby increasing the autocorrelations be-
tween the members of the time series. Here, following
Chen (1982), we estimate the effective time between in-
dependent members t, and the effective degrees of free-
dom, using the autoregressive properties of both time
series. All the correlations on interdecadal time scale
discussed in the following sections are statistically
checked for significance at 95% confidence level, using
the corresponding effective degrees of freedom com-
puted by this method.
Figure 5a shows the 11-yr running mean of the sea-
sonal mean KE and the exchange at 200 hPa calculated
using ERA-40. The low-frequency seasonal mean KE
exhibits a linear decreasing trend with a multidecadal
variability embedded in it. The low-frequency rate of
KE exchange between the seasonal mean and ISO is
always positive except for some part of 1960s, and it
shows a linear increasing trend and multidecadal vari-
ability. The seasonal mean KE and the exchange are
strongly linked with a correlation coefficient of 20.71
(Table 1). The relationship is unaffected whether the
linear trend is removed from both time series or not.
Hence, the internal processes can explain nearly half
of the interdecadal ISM variability. The weakening of
upper-level ISM strength and its multidecadal variability
can be explained in terms of energy exchange between
the seasonal mean and ISO. During the initial periods,
when seasonal mean KE was high, ISO supplied KE to
the seasonal mean but later the seasonal mean started
FIG. 5. Low-pass filtered seasonal mean KE per unit mass (m2 s22, solid line) and the rate of KE (per unit mass) exchange between the
seasonal mean and 10–60-day time scale (31025 W kg21, dashed line) at 200 hPa calculated using ERA-40 data. (a) Unaltered variability;
(b) variability without ENSO; (c) variability without PDO; and (d) variability without AMO. Variability associated with ENSO, PDO,
and AMO is removed from both seasonal mean KE and seasonal mean and 10–60-day KE exchange. The corresponding correlation
coefficients (CC) are given at the upper-left corner of each panel.
1768 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69
Page 9
losing energy to the ISO. The fluctuations in the rate of
KE exchange can account for the local maxima and
minima in the seasonal mean KE. It can be deduced that
ISO takes more and more energy from the seasonal
mean and makes it lose its strength. Still, the question of
modulation of interdecadal energy exchange process by
external forcing may arise. The IDV of ENSO, IOD,
PDO, and AMO was brought out by applying a 11-yr
running mean on the JJAS mean Nino-3.4, IOD, PDO,
and North Atlantic SST anomaly indices, respectively.
Unlike the case of the interannual time scale, on the in-
terdecadal scale the exchange process and external forc-
ings are linked (Table 1). Figures 5b–d show the ENSO-,
PDO-, and AMO-removed interdecadal seasonal mean
KE and the rate of KE exchange between seasonal mean
and ISO. The removal of external forcings not only pre-
served the linear trend and multidecadal variability in the
seasonal mean KE and the exchange process but also
improved the correlation coefficient between them
(Table 1). It may be cautioned that in our analysis we
removed only the linear relationship between the ex-
change and external forcing. There may still be some
variability associated with complex nonlinear interaction
among the different external forcing components.
The internal interdecadal variability is further exam-
ined at the lower level by applying an 11-yr running
mean on seasonal mean KE and the energy exchange
process at 850 hPa. Contrary to the observation of linear
decreasing trend in the upper-atmospheric ISM strength,
lower-atmospheric seasonal mean KE exhibits a linear
increasing trend on interdecadal time scales (Fig. 6). It
also shows a multidecadal variability. The exchange
process is always negative; that is, the ISO always supplies
energy to the seasonal mean. Analogous to the upper
atmosphere, the long-term behavior and fluctuations
can be explained in terms of energy exchange. The 11-yr
running mean smooths out the high-frequency fluctua-
tions and reveals a strong link between strength of the
ISM and the energy exchange process on the interdecadal
time scale. The relationship is unaltered whether the
known quantifiable external forcing signal is removed
or not. Consistent with the upper-atmospheric obser-
vation, the internal process can explain about 50% of
the interdecadal variability of the ISM. In the next sec-
tion, we discuss the sensitivity of our results to the tem-
poral window length.
6. Temporal window length and the estimate oftime mean–ISO interaction
Since the number of resolvable harmonics of a time
series depends on the length of the time series, it is
possible that the estimate of seasonal mean–ISO energy
exchange may be influenced by the temporal window
length. If the energy exchange is very much sensitive to
the temporal window length, the estimation of IAV of
seasonal mean due to the seasonal mean–ISO energy
exchange may lead to unrealistic conclusions. To check
the robustness of our results, we repeated our calculation
by extending the length of the time series from 122 to 150
days, considering the data from 17 May to 14 October.
The newly constructed time series not only brings in more
harmonics in the ISO time scale but also extends the
largest resolvable time scale from 60 to 75 days. Figure 7
compares the seasonal mean–ISO energy exchange esti-
mated using 122- and 150-day windows at the 200- and
850-hPa pressure levels. Consistency in the estimation of
energy exchange is observed at both levels at a correla-
tion of 0.88 and 0.83, respectively (Figs. 7a,b), and it is
reflected in the correlation coefficient of IAV of seasonal
mean KE and the energy exchange between seasonal
mean and 10–75-day KE (20.47 at 200 hPa and 20.41 at
850 hPa). The inclusion of the 75-day scale and higher
time scale resolution in the estimation of IAV of the
seasonal mean due to feedback between seasonal mean
and ISO brings about only a slight change, and the dis-
crepancy is not significant.
7. Reanalysis data and rate of KE exchangecalculation
The reliability of our results depends upon how well
ERA-40 data capture the behavior of the true atmo-
sphere. To quantify the interannual and interdecadal
variability, long datasets are required. The changes in
observing systems at the beginning of satellite era after
the mid-1970s always raise a concern about the use of
reanalysis products for analyzing long-term trends and
multidecadal variability (Bengtsson et al. 2004). The
consistency of our results is checked by repeating the
analysis at 200 hPa with another independent dataset,
FIG. 6. Low-pass filtered seasonal mean kinetic energy per unit
mass (m2 s22, solid line) and the rate of kinetic energy (per unit
mass) exchange between seasonal mean and 10–60-day time scale
(31025 W kg21, dashed line) at 850 hPa calculated using ERA-40
data. The correlation coefficient (CC) is given at the upper-left
corner of the figure.
JUNE 2012 S U H A S E T A L . 1769
Page 10
NCEP–NCAR reanalysis data, for the 1958–2001 JJAS
months, and it is presented in Fig. 8. Consistent with the
result obtained from the ERA-40 data, it also shows
a moderate correlation of 20.42, which is statistically
significant at the 95% confidence level.
Although the amount of variability explained by the
internal process is the same for both reanalysis products,
the decreasing trend of seasonal mean KE at 200 hPa
was more pronounced in the ERA-40 dataset. In addi-
tion, the nature of variability of the rate of KE exchange
was different for both datasets. It can be noted from
Figs. 9a and 9b that although the interannual seasonal
mean KEs are in phase, NCEP–NCAR data overestimate
the seasonal mean KE at 200 and 850 hPa. To examine
this discrepancy, the seasonal mean KE is disintegrated
into zonal and meridional components at 850 and
200 hPa and are presented in Fig. 10. It shows that there
are differences in both amplitude and phases of meridi-
onal component of the seasonal mean KE at both levels.
Also, there is no phase relationship between the zonal
and meridional components of the seasonal mean KE of
NCEP–NCAR dataset. For example, at 200 hPa, sea-
sonal mean U2/2 is decreasing but the seasonal mean V2/2
shows a steady linear increasing trend. The cross-spectral
method of estimation of KE exchange involves products
of spectral coefficient and seasonal mean wind com-
ponents. Hence, the difference in the phases of wind
components can cause a discrepancy in the exchange
estimation. The difference in the meridional KE be-
tween the two reanalyses may be related to the fact that
the tropical divergent component of the wind is rather
weak in the NCEP–NCAR reanalysis (Annamalai et al.
1999). Because of the biases in the meridional component
of NCEP–NCAR reanalysis wind data, throughout our
study we have used ERA-40 data, which show a coherent
behavior of zonal and meridional wind at both 850 and
200 hPa.
8. Summary and conclusions
In this study, we have introduced a conceptual
framework for quantifying the role of ISO in causing
FIG. 7. Rate of KE per unit mass (31025 W kg21) exchange
between seasonal mean and 10–60-day time scale (solid line) and
rate of KE per unit mass (31025 W kg21) exchange between
seasonal mean and 10–75-day time scale (dashed line) calculated
using ERA-40 data (a) at 200 hPa over the region 08–158N, 508–808E
and (b) at 850 hPa over the region 08–208N, 508–708E. The corre-
sponding correlation coefficients (CC) are given at the upper-left
corner of each panel.
FIG. 8. Seasonal mean kinetic energy per unit mass (m2 s22, solid
line) and the rate of kinetic energy (per unit mass) exchange be-
tween seasonal mean and 10–60-day time scale (31025 W kg21,
dashed line) at 200 hPa calculated using NCEP–NCAR reanalysis
data from 1958 to 2001. The correlation coefficient (CC) is given at
the upper-left corner of the figure.
FIG. 9. Seasonal mean (JJAS) kinetic energy per unit mass
(m2 s22) calculated using NCEP–NCAR reanalysis (solid line) and
ERA-40 data (dashed line) at (a) 200 hPa averaged over 08–158N,
508–808E and (b) 850 hPa averaged over 08–208N, 508–708E.
1770 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69
Page 11
interannual and interdecadal variability of the seasonal
mean ISM. Although there have been some studies that
indicated evidence for such an interaction over the ISM
domain, such as the correlation between seasonal mean
and ISO variance, a clear mechanism through which
such an interaction takes place had remained elusive
(Goswami and Ajaya Mohan 2001; Lawrence and
Webster 2001; Qi et al. 2008). As the ISOs represent
a dominant mode of summer monsoon variability, we
hypothesize that the ISOs play a significant role in
modulating the seasonal mean through scale interaction
and causing internally generated interannual and inter-
decadal variability of the seasonal mean.
A cross-spectral method of estimation of KE ex-
change in the frequency domain (Hayashi 1980) is em-
ployed for quantifying the interannual and interdecadal
seasonal mean–ISO interaction. The analysis is done
using ERA-40 wind data at the 850- and 200-hPa pres-
sure levels over the region where the seasonal mean KE
shows a significant relationship with the seasonal mean
ISM rainfall. It is found from the energy exchange cal-
culation in both the upper and lower atmosphere that
the amplitude and phases of the net rate of KE exchange
between seasonal mean and ISO is determined by the
convergence of ISO KE by seasonal mean zonal wind.
The internal interannual and interdecadal variabilities
were quantified by correlating the seasonal mean KE
phase variability with the rate of KE exchange between
the seasonal mean and ISO at upper and lower levels. It
is found that the long-term trend and much of the vari-
ability of the seasonal mean KE could be related to the
exchange of energy between seasonal mean and ISO.
The issue of sensitivity to choice of temporal window
length on the energy exchange is addressed separately
by extending the length of the time series. The KE en-
ergy exchange process at 200 hPa is less noisy and rel-
atively free of boundary layer turbulence, hence its
variability with the seasonal mean KE is considered for
the quantification of internal variability of the ISM
seasonal mean. Although contaminated by friction and
boundary flux terms, the lower-level atmosphere also
shows a statistically significant relationship, which is also
independent of any known and quantifiable external
boundary forcing. On the whole, the influence of ex-
ternal forcing on the exchange process is negligible on
interannual time scale, while on interdecadal time scale
external forcing does have some contribution. The in-
ternal IAV of the ISM could independently account for
approximately 20% of the ISM seasonal mean vari-
ability. It is shown that in both the upper and lower at-
mosphere the energy exchange process could explain
about 50% of the variability of the ISM seasonal mean
on an interdecadal time scale. This conclusion was
drawn on the assumption that the relationship between
the exchange and the different external forcing compo-
nents is linear. There may still exist some unaccounted
external variability arising from nonlinearity and this may
be a caveat in our analysis. Another issue that demands
FIG. 10. (a),(c) Seasonal mean (JJAS) U2/2 (m2 s22) calculated using NCEP–NCAR reanalysis (solid line) and ERA-40 data (dashed
line), at (a) 200 and (c) 850 hPa. (b),(d) Seasonal mean (JJAS) V2/2 (m2 s22) calculated using NCEP–NCAR reanalysis (solid line) and
ERA-40 data (dashed line) at (b) 200 and (d) 850 hPa.
JUNE 2012 S U H A S E T A L . 1771
Page 12
a closer examination is the weakening of the relationship
between seasonal mean and ISOs after the mid-1970s. At
present, it is not clear whether this difference is due to the
reported mid-1970s climate regime shift or is simply an
artifact of reanalysis dataset.
The weakening of the monsoon circulation on inter-
decadal time scales as evidenced by a decreasing trend of
seasonal mean KE of the TEJ is often attributed to global
warming (Vecchi et al. 2006; Vecchi and Soden 2007).
Based on a modeling study, Chou and Chen (2010) pro-
posed that increase in global temperature would cause
an uplifting of the tropopause, which may favor deeper
convection and eventually lead to an increase in the sta-
bility and weakening of the tropical circulation. However,
we speculate that this decreasing trend of seasonal mean
KE could be partially explained through an internal dy-
namical process where the mean is losing increasingly
more energy to the ISOs through scale interactions.
Concerns about the usage of reanalysis data for the
study of multidecadal variability and long-term trend
are also discussed separately. It is found that the zonal
component of KE is slightly overestimated in NCEP–
NCAR reanalysis but is well correlated on interannual
time scale with that in the ERA-40 dataset. However,
there are some notable differences in the meridional
component of KE in the two products. Significant re-
lationships between the ERA-40 wind data and rain
gauge data add further confidence in using it. It is already
known that the ISO can influence the synoptic activity
through clustering of lows and depression along the
monsoon trough (Goswami et al. 2003). In the present
study, we demonstrate the role of ISO in modulating
the ISM seasonal mean on interannual and interdecadal
time scale. This emphasizes on the need for further
research in improving the simulation of MISO in order
to improve the prediction of monsoon on short, me-
dium, and longer time scales, since it not only affects
active and break spells but also affects the variability in
other scales.
Acknowledgments. The Indian Institute of Tropical
Meteorology is funded by the Ministry of Earth Sci-
ences, Government of India. SE and NJM are grateful to
CSIR for a Fellowship.
APPENDIX
Derivation of Rate of Change of Time MeanKinetic Energy
The rate of change of time mean kinetic energy in the
transient eddy–time mean framework is represented by
Eq. (1). The equation is derived in Cartesian coordinates
from the momentum equations and the continuity
equation by assuming that the large-scale atmosphere
is in the state of hydrostatic balance. Variant forms of
Eq. (1) were used by different researchers (Holopainen
1978b; Lau and Lau 1992; Maloney and Hartmann
2001, Serra et al. 2008) to quantify the eddy–time mean
interactions. Here we again present Eq. (1):
d(K)/dt 5 BC 1 CAK 1 WF 1 D.
Variables with an overbar represent the time mean and
the primed quantities represent the transient eddies. The
expansion of the components of Eq. (1) is given below:
BC 5 u9u9›u
›x1 u9y9
›u
›y1
›y
›x
� �1 u9y9
›y
›y, (A1)
CAK 5 2av, (A2)
WF 5 $ � [fV 1 k � V 1 (V � V9)V9], (A3)
D 5 V � f , (A4)
where V represents the horizontal velocity and u, y, and
v respectively represent the zonal, meridional, and ver-
tical components of velocity. Also, k stands for kinetic
energy per unit mass [i.e., (u 3 u 1 y 3 y)/2], F and a
represent the geopotential height and specific volume,
respectively, and f stands for friction.
REFERENCES
Ajaya Mohan, R. S., and B. N. Goswami, 2003: Potential pre-
dictability of the Asian summer monsoon on monthly and sea-
sonal time scales. Meteor. Atmos. Phys., 84, 83–100, doi:10.1007/
s00703-002-0576-4.
Annamalai, H., J. Slingo, K. R. Sperber, and K. Hodges, 1999: The
mean evolution and variability of the Asian summer monsoon:
Comparison of ECMWF and NCEP–NCAR analysis. Mon.
Wea. Rev., 127, 1157–1186.
Awade, S. T., M. Y. Totagi, and S. M. Bawiskar, 1982: Wave to
wave and wave to zonal mean flow kinetic energy exchanges
during contrasting monsoon years. Pure Appl. Geophys., 120,
463–482.
Bengtsson, L., S. Hagemann, and K. I. Hodges, 2004: Can climate
trends be calculated from reanalysis data? J. Geophys. Res.,
109, D11111, doi:10.1029/2004JD004536.
Charney, J. G., and J. Shukla, 1981: Predictability of monsoons.
Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cam-
bridge University Press, 99–109.
Chen, W. Y., 1982: Fluctuations in Northern Hemisphere 700-mb
height field associated with the Southern Oscillation. Mon.
Wea. Rev., 110, 808–823.
Chou, C., and C.-A. Chen, 2010: Depth of convection and the
weakening of tropical circulation in global warming. J. Cli-
mate, 23, 3019–3030.
1772 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69
Page 13
Ferranti, L., J. M. Slingo, T. N. Palmer, and B. J. Hoskins, 1997:
Relations between interannual and intraseasonal monsoon var-
iability as diagnosed from AMIP integrations. Quart. J. Roy.
Meteor. Soc., 123, 1323–1357.
Fujinami, H., and Coauthors, 2011: Characteristic intraseasonal
oscillation of rainfall and its effect on interannual variability
over Bangladesh during boreal summer. Int. J. Climatol., 31,
1192–1204, doi:10.1002/joc.2146.
Gadgil, S., M. Rajeevan, and R. Nanjundiah, 2005: Monsoon
prediction? Why yet another failure? Curr. Sci., 88, 1389–
1400.
Goswami, B. N., 1995: A multiscale interaction model for the origin
of the tropospheric QBO. J. Climate, 8, 524–534.
——, 1998: Interannual variation of Indian summer monsoon in
a GCM: External conditions versus internal feedbacks. J. Cli-
mate, 11, 501–522.
——, 2005: South Asian monsoon. Intraseasonal Variability of the
Atmosphere–Ocean Climate System, W. K. M. Lau and D. E.
Waliser, Eds., Springer, 19–61.
——, and R. S. Ajaya Mohan, 2001: Intraseasonal oscillations
and inter-annual variability of the Indian summer monsoon.
J. Climate, 14, 1180–1198.
——, and P. K. Xavier, 2005: Dynamics of ‘‘internal’’ interannual
variability of Indian summer monsoon in a GCM. J. Geophys.
Res., 110, D24104, doi:10.1029/2005JD006042.
——, R. S. Ajaya Mohan, P. K. Xavier, and D. Sengupta, 2003:
Clustering of low pressure systems during the Indian summer
monsoon by intraseasonal oscillations. Geophys. Res. Lett., 30,
1431, doi:10.1029/2002GL016734.
——, M. S. Madhusoodanan, C. P. Neema, and D. Sengupta, 2006a:
A physical mechanism for North Atlantic SST influence on the
Indian summer monsoon. Geophys. Res. Lett., 33, L02706,
doi:10.1029/2005GL024803.
——, G. Wu, and T. Yasunari, 2006b: Annual cycle, intraseasonal
oscillations and roadblock to seasonal predictability of the
Asian summer monsoon. J. Climate, 19, 5078–5099.
Hahn, D. J., and J. Shukla, 1976: An apparent relationship between
Eurasian snow cover and Indian monsoon rainfall. J. Atmos.
Sci., 33, 2461–2462.
Hayashi, Y., 1980: Estimation of nonlinear energy transfer spectra
by the cross-spectral method. J. Atmos. Sci., 37, 299–307.
Held, I. M., and A. Y. Hou, 1980: Nonlinear axially symmetric
circulations in a nearly inviscid atmosphere. J. Atmos. Sci., 37,
515–533.
Holopainen, E. O., 1978a: On the dynamical forcing of the long-
term mean flow by the large-scale Reynolds’ stresses in the
atmosphere. J. Atmos. Sci., 35, 1596–1604.
——, 1978b: A diagnostic study on the kinetic energy balance of the
long-term mean flow and the associated transient fluctuations
in the atmosphere. Geophysica, 15, 125–145.
Hoskins, B. J., and R. P. Pearce, 1983: Large-Scale Dynamical
Processes in the Atmosphere. Academic Press, 397 pp.
——, I. N. James, and G. H. White, 1983: The shape, propaga-
tion, and mean flow interaction of large-scale weather systems.
J. Atmos. Sci., 40, 1595–1612.
Hoyos, C. D., and P. J. Webster, 2007: The role of intraseasonal
variability in the nature of Asian monsoon precipitation.
J. Climate, 20, 4402–4424.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Re-
analysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.
Kanamitsu, M., T. N. Krishnamurti, and C. Depradine, 1972: On
scale interactions in the tropics during northern summer.
J. Atmos. Sci., 29, 698–706.
Kang, I.-S., and J. Shukla, 2006: Dynamic seasonal prediction and
predictability. The Asian Monsoon, Springer, 585–612.
——, J. Y. Lee, and C. K. Park, 2004: Potential predictability of
summer mean precipitation in a dynamical seasonal prediction
system with systematic error correction. J. Climate, 17, 834–
844.
Krishna Kumar, K., M. Hoerling, and B. Rajagopalan, 2005: Ad-
vancing dynamical prediction of Indian monsoon rainfall.
Geophys. Res. Lett., 32, L08704, doi:10.1029/2004GL021979.
——, B. Rajagopalan, M. Hoerling, G. Bates, and M. Cane, 2006:
Unraveling the mystery of Indian monsoon failure during
El Nino. Science, 314, 115–119, doi:10.1126/science.1131152.
Krishnamurthy, V., and B. N. Goswami, 2000: Indian monsoon–
ENSO relationship on interdecadal time scales. J. Climate, 13,
579–595.
——, and J. Shukla, 2008: Seasonal persistence and propagation of
intraseasonal patterns over the Indian monsoon region. Cli-
mate Dyn., 30, 353–369.
Krishnamurti, T. N., and D. R. Chakraborty, 2005: The dynamics of
phase locking. J. Atmos. Sci., 62, 2952–2964.
——, D. Bachiochi, T. Larow, B. Jha, M. Tewari, D. R. Chakraborty,
R. Correa-Torres, and D. Oosterhof, 2000: Coupled atmosphere–
ocean modeling of the El Nino of 1997/98. J. Climate, 13,
2428–2459.
——, D. R. Chakraborty, N. Cubukcu, L. Stefanova, and T. S. V.
Vijaya Kumar, 2003: A mechanism of the Madden–Julian
oscillation based on interactions in the frequency domain.
Quart. J. Roy. Meteor. Soc., 129, 2559–2590.
Lau, K.-H., and N.-C. Lau, 1992: The energetic and propagation
dynamics of tropical summer time synoptic-scale disturbances.
Mon. Wea. Rev., 120, 2523–2539.
Lawrence, D. M., and P. J. Webster, 2001: Interannual variations of
the intraseasonal oscillation in the South Asian summer
monsoon region. J. Climate, 14, 2910–2922.
Maloney, E. D., and D. L. Hartmann, 2001: The Madden–Julian
oscillation, barotropic dynamics, and North Pacific tropical
cyclone formation. Part I: Observations. J. Atmos. Sci., 58,
2545–2558.
Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C.
Francis, 1997: A Pacific interdecadal climate oscillation with
impacts on salmon production. Bull. Amer. Meteor. Soc., 78,
1069–1079.
Neena, J. M., and B. N. Goswami, 2010: Extension of potential
predictability of Indian summer monsoon dry and wet spells in
recent decades. Quart. J. Roy. Meteor. Soc., 136, 583–592.
Qi, Y., R. Zhang, T. Li, and M. Wen, 2008: Interactions between
the summer mean monsoon and the intraseasonal oscillation
in the Indian monsoon region. Geophys. Res. Lett., 35, L17704,
doi:10.1029/2008GL034517.
Rajeevan, M., J. Bhate, and A. K. Jaswal, 2008: Analysis of vari-
ability and trends of extreme rainfall events over India using
104 years of gridded daily rainfall data. Geophys. Res. Lett., 35,
L18707, doi:10.1029/2008GL035143.
Rasmusson, E. M., and T. H. Carpenter, 1983: The relationship
between eastern equatorial Pacific sea surface temperatures and
rainfall over India and Sri Lanka. Mon. Wea. Rev., 111, 517–528.
Saltzman, B., 1957: Equations governing the energetics of the
larger scales of atmospheric turbulence in the domain of wave
number. J. Atmos. Sci., 14, 513–523.
Schneider, E. K., and R. S. Lindzen, 1977: Axially symmetric
steady-state models of the basic state for instability and cli-
mate studies. Part I. Linearized calculations. J. Atmos. Sci., 34,
263–279.
JUNE 2012 S U H A S E T A L . 1773
Page 14
Seiki, A., and Y. N. Takayabu, 2007: Westerly wind bursts and their
relationship with intraseasonal variations and ENSO. Part II:
Energetics over the western and central Pacific. Mon. Wea.
Rev., 135, 3346–3361.
Serra, Y. L., G. N. Kiladis, and M. F. Cronin, 2008: Horizontal and
vertical structure of easterly waves in the Pacific ITCZ.
J. Atmos. Sci., 65, 1266–1284.
Sheng, J., and Y. Hayashi, 1990: Estimation of atmospheric ener-
getics in the frequency domain during the FGGE year. J. At-
mos. Sci., 47, 1255–1268.
Shukla, J., 1998: Predictability in the midst of chaos: A scientific
basis for climate forecasting. Science, 282, 728–731.
——, and D. Paolino, 1983: The Southern Oscillation and the long-
range forecasting of the summer monsoon rainfall over India.
Mon. Wea. Rev., 111, 1830–1837.
Sperber, K. R., J. M. Slingo, and H. Annamalai, 2000: Predictability
and the relationship between subseasonal and interannual
variability during the Asian summer monsoon. Quart. J. Roy.
Meteor. Soc., 126, 2545–2574.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis.
Quart. J. Roy. Meteor. Soc., 131, 2961–3012.
Vecchi, G. A., and B. J. Soden, 2007: Global warming and the
weakening of the tropical circulation. J. Climate, 20, 4316–4340.
——, ——, A. T. Wittenberg, I. M. Held, A. Leetmaa, and M. J.
Harrison, 2006: Weakening of tropical Pacific atmospheric
circulation due to anthropogenic forcing. Nature, 441, 73–76.
Webster, P. J., V. O. Magana, T. N. Palmer, J. Shukla, R. A. Tomas,
M. Yanai, and T. Yasunari, 1998: Monsoons: Processes, pre-
dictability, and the prospects for prediction. J. Geophys. Res.,
103, 14 451–14 510.
1774 J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S VOLUME 69