Interactive Feedback between ENSO and the Indian Ocean
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
1784 JOURNAL OF CLIMATE VOLUME 19
Interactive Feedback between ENSO and the Indian Ocean
JONG-SEONG KUG AND IN-SIK KANG
Climate Environment System Research Center, Seoul National University, Seoul, South Korea
(Manuscript received 1 April 2004, in final form 10 August 2005)
ABSTRACT
A feedback process of the Indian Ocean SST on ENSO is investigated by using observed data and
atmospheric GCM. It is suggested that warming in the Indian Ocean produces an easterly wind stress
anomaly over Indonesia and the western edge of the Pacific during the mature phase of El Niño. The
anomalous easterly wind in the western Pacific during El Niño helps a rapid termination of El Niño and a
fast transition to La Niña by generating upwelling Kelvin waves. Thus, warming in the Indian Ocean, which
is a part of the El Niño signal, operates as a negative feedback mechanism to ENSO. This Indian Ocean
feedback appears to operate mostly for relatively strong El Niños and results in a La Niña one year after
the mature phase of the El Niño. This 1-yr period of phase transition implies a possible role of Indian
Ocean–ENSO coupling in the biennial tendency of the ENSO. Atmospheric GCM experiments show that
Indian Ocean SST forcing is mostly responsible for the easterly wind anomalies in the western Pacific.
1. Introduction Kelvin waves, which play a role in the ENSO transition.
However, whether the mechanisms, suggested by pre-
Since Bjerknes (1969), considerable research effort
vious studies, can fully explain wind anomalies in the
has been devoted to gaining a better understanding of
western Pacific is being disputed.
ENSO dynamics. In the accumulating of our under-
In terms of causing wind anomalies over the western
standing of the ENSO, various conceptual theories
Pacific, Weisberg and Wang (1997a) and their subse-
were suggested to explain the self-sustained oscillation
quent studies emphasized a role of local cold SST
of the ENSO. Among them, the delay oscillator
anomalies induced by Ekman pumping. According to
(Schopf and Suarez 1988; Battisti and Hirst 1989) and
their argument, the El Niño–related anomalous convec-
the recharge oscillator (Jin 1997a,b) are two of the most
tion in the equatorial central Pacific induces a pair of
popular theories. Both theories explain the developing
off-equatorial cyclones with westerly wind anomalies
mechanism of ENSO in a similar manner but suggest
on the equator. These equatorial westerly winds act to
explicitly different phase transition mechanisms. How-
deepen the thermocline and increase SST in the equa-
ever, both emphasize basinwide ocean adjustment pro-
torial east-central Pacific, thereby providing a positive
cesses in nonequilibrium with wind stress forcing and
feedback for anomaly growth. At the same time, these
the role of oceanic waves in the tropical Pacific basin
off-equatorial cyclones raise the off-equatorial ther-
(An and Kang 2001).
mocline by Ekman pumping. Thus, a relatively shallow
On the other hand, several recent studies have fo-
off-equatorial thermocline anomaly develops over the
cused on a role for the western Pacific wind in the
western Pacific, leading to a decrease in SST and an
ENSO transition (Weisberg and Wang 1997a,b; C.
increase in sea level pressure (SLP). During the mature
Wang et al. 1999; B.Wang et al. 1999, 2001). They em-
phase of El Niño, this off-equatorial high SLP initiates
phasized a role for anomalous easterly (westerly) winds
equatorial easterly winds over the far western Pacific.
in the western Pacific during the mature phase of El
Wang and his collaborators (B. Wang et al. 1999,
Niño (La Niña). The wind anomalies excite equatorial
2000, 2001; Wang and Zhang 2002) have suggested that
the easterly wind anomalies are associated with the
anomalous Philippine Sea anticyclone. In contrast to
Corresponding author address: Prof. In-Sik Kang, School of
Earth and Environmental Sciences, Seoul National University, Weisberg and Wang (1997a,b), they emphasized the
Seoul, 151-742, South Korea. role of local air–sea interaction. The anomalous anticy-
E-mail: kang@climate.snu.ac.kr clone winds superposed on mean northeasterly trades
© 2006 American Meteorological Society
JCLI36601 MAY 2006 KUG AND KANG 1785
during the boreal wintertime increase the total wind in the decoupled run. These controversial results
speed, thus enhancing evaporation and entrainment to should be further studied using observational data and
the east of the anticyclone. This favors sea surface cool- experimental modeling.
ing in front of the anticyclone. Thus, the cold SST is In this study, we will show that the Indian Ocean SST
located to the east of the anticyclone. The phase shift anomalies affect ENSO variability using observational
between the anticyclone and sea surface cooling implies data. We will show that this coupling process is gener-
a local feedback that can amplify and sustain the Phil- ated by a modulating Walker circulation and altering
ippine Sea anticyclonic wind. The anomalous equato- wind variability over the western Pacific. We will also
rial easterly wind can be regarded as a part of the an- demonstrate that the anomalous easterly wind in the
ticyclone. western Pacific during El Niño helps a fast transition to
Previous studies have recognized that the easterly La Niña by generating upwelling Kelvin waves.
winds are caused by the off-equatorial cold SSTs and Section 2 introduces the observed data and the model
their resultant anticyclonic flow. However, there is in- utilized. In section 3, we show a clear relationship be-
creasing research interest on covariability between the tween ENSO and the Indian Ocean SST variability us-
Indian Ocean and the western Pacific (e.g., Li et al. ing observed data, and the dynamical coupling process
2002; Watanabe and Jin 2002, 2003; Wang et al. 2003; between ENSO and the Indian Ocean will be de-
Misra 2004; Kug et al. 2005). In particular, Watanabe scribed. Atmospheric GCM experiments support our
and Jin (2002) showed, using their moist linear baro- arguments in section 4. The summary and discussion
clinic model, that the Indian Ocean SST plays a role in are given in section 5.
developing the Philippine Sea anticyclone. In a similar
manner, we will suggest that the easterly anomalies in
2. Data and model
the equatorial western Pacific are closely related to In-
dian Ocean warming. In addition, we will emphasize The data utilized are monthly means of sea surface
that the Indian Ocean SST plays a crucial role in the temperature (SST), atmospheric circulation data, and
tropical Pacific ENSO variability. ocean temperature over the tropical domain, 30°S–
So far, a number of studies have focused on the im- 30°N. The data period is 43 yr from 1958 to 2000.
pact of ENSO on the Indian Ocean (e.g., Klein et al. Anomalies presented in this study were detrended after
1999; Venzke et al. 2000; Baquero-Bernal et al. 2002; removing climatology for 43 yr. The SST data were
Xie et al. 2002; Lau and Nath 2000, 2003; Krishnamur- obtained from the National Centers for Environmental
thy and Kirtman 2003). To a large extent, they agreed Prediction (NCEP) and were constructed based on the
that the ENSO-related Indian Ocean SST variability is EOF of observed SST (Reynolds 1988) and recon-
mostly affected by modulation of the Walker circula- structed after January 1981 using the optimum interpo-
tion. However, little research has been devoted to the lation technique (Reynolds and Smith 1994). Atmo-
impact of the Indian Ocean SST on the ENSO. During spheric circulation data were taken from NCEP–
a few recent years, some studies have suggested that the National Center for Atmospheric Research (NCAR;
Indian Ocean variability can modulate the ENSO vari- Kalnay et al. 1996). They have a horizontal resolution
ability (e.g., Yu et al. 2002; Saji and Yamagata 2003; Wu of 2.5° ⫻ 2.5°.
and Kirtman 2004; Annamalai et al. 2005; Kug et al. The ocean temperature data utilized are the monthly
2005). In particular, Yu et al. (2002) and Wu and Kirt- means of the Simple Ocean Data Assimilation (SODA)
man (2004) examined the impacts of the Indian Ocean product (Carton et al. 2000a,b). Xie et al. (2002) show
on the ENSO variability by performing experiments that the SODA data have a good agreement with the
with a coupled atmosphere–ocean general circulation expendable bathythermograph (XBT). It is available at
model (GCM), respectively. Both studies showed that a 1° ⫻ 1° resolution in the midlatitudes and 0.45° ⫻ 1°
coupled GCM simulation of ENSO including both the latitude–longitude resolution in the Tropics, and has 20
tropical Indian and Pacific Oceans tends to be a stron- vertical levels with 15-m resolutions near the sea sur-
ger ENSO variability than that including the tropical face.
Pacific only. However, they gave opposite results in An atmospheric GCM is used to reproduce the ob-
terms of the dominant ENSO period. Yu et al. (2002) served characteristics of atmospheric responses to given
showed the dominant period of the ENSO increases SST anomalies. The GCM is a spectral model with a
from about 4 yr in the Pacific run to about 4.4 yr in the triangular truncation at wavenumber 42 and has 20 ver-
Indo-Pacific run, while Wu and Kirtman (2004) showed tical levels. The model was originally developed at the
that the dominant ENSO period decreases from about University of Tokyo and modified at Seoul National
2.3 yr in the Indian Ocean coupled run to about 2.8 yr University (SNU). The physical processes included are1786 JOURNAL OF CLIMATE VOLUME 19
FIG. 1. Lag-correlation coefficients between the WP zonal wind index during November–
January and equatorial SSTs (5°S–5°N). The y axis denotes the calendar months of the ENSO
developing year (year 0) and decaying year (year 1). Light, medium, and dark shading rep-
resent the 90%, 95%, and 99% confidence levels, respectively.
the Nakajima two-stream scheme for longwave and Figure 1 shows lag-correlation coefficients between
shortwave radiation (Nakajima et al. 1995), the Re- the zonal wind index and equatorial SST averaged over
laxed Arakawa–Shubert scheme (Moorthi and Suarez 5°S–5°N during the years 1958–2001. As reported from
1992), shallow convection, land surface processes, grav- previous studies (e.g., C. Wang et al. 1999; Wang et al.
ity wave drag, and planetary boundary layer processes 2001), the zonal wind near the equatorial western
(Kim et al. 1998). Kim et al. (1998) showed that the boundary is simultaneously correlated to the eastern
SNU GCM reasonably simulates the climatological Pacific SST. The correlation coefficients are greater
mean patterns of tropical circulation and their anoma- than 0.4, having a 95% confidence level. In addition,
lies during El Niño. the wind index is correlated to the local western Pacific
SST with a 95% confidence level as pointed out by
3. Observational analysis Weisberg and Wang (1997a,b). However, Fig. 1 also
shows that the wind index is well correlated to the SST
As mentioned in the introduction, the equatorial in the western Indian Ocean with 1–2-month lag. The
wind anomalies in the western Pacific may be important correlation coefficients are greater than 0.6, having a
for decaying the eastern Pacific SST anomalies associ- 99% confidence level. This means that the establish-
ated with El Niño and La Niña. To examine what is of ment of the equatorial zonal wind is more closely re-
relevance to the zonal wind anomalies, a western Pa- lated to the Indian Ocean SST forcing rather than the
cific (WP) zonal wind index is defined as a 1000-hPa eastern Pacific SST and the western Pacific SST.
zonal wind area averaged over 5°S–5°N, 120°–150°E It is also interesting that the WP zonal wind index is
from November to January when the typical ENSO well correlated with the next year’s eastern Pacific SST
events are in their mature phase. The WP wind index with a 99% confidence level. That is, the easterly wind
represents the strength of the zonal wind near the equa- anomaly near the western boundary during an El Niño
torial western Pacific boundary. In fact, the index re- mature phase leads a rapid termination of the El Niño
gion includes land and inland sea, where the wind can- and sequentially fast phase transition to La Niña within
not influence ENSO variability. It seems to be appro- 1 yr, judging the time scale of the typical ENSO as
priate to remove the land and inland sea region when being 3–5 yr. One can explain the relation as follows:
the WP wind index is calculated. However, we found during an El Niño mature phase, the easterly wind
that the time series with and without masking out those anomalies generate upwelling Kelvin waves, which
regions are very close, as the correlation coefficient be- propagate eastward, and accelerate to break up the
tween them is 0.97. Therefore, for simplicity we use the warm SST. Therefore, the transition from El Niño to La
time series without masking out the land and inland sea Niña is progressed rapidly within 1 yr. This procedure
regions in this study. will be shown later in this section.1 MAY 2006 KUG AND KANG 1787
FIG. 2. Time series of the Niño-3.4 SST during November–January (bar) and WISSTON
(triangle).
To show the role of the Indian Ocean SST on ENSO All positive WISST events coincided with El Niño
clearly, we defined a western Indian Ocean SST index events. Following the El Niño events, the phase transi-
(WISSTON),1 area averaged over 10°S–10°N, 55°–75°E, tion progressed rapidly, and La Niña develops in the
where the correlation with WP zonal wind index is rela- winter of the following year. We checked that this re-
tively high during October–November. Figure 2 shows lationship is very robust for the long historical SST
the time series of the Niño-3.4 SST and WISSTON. The (1885–2000). For this period, there were 16 warm west-
Niño-3.4 SST is averaged over 5°S–5°N, 170°–120°W ern Indian Ocean cases, more than its one standard
during November–January. The correlation coefficient deviation. Among them, 13 cases led to the La Niña
between two indices is 0.62. The high correlation is not state in the next year (not shown).
an unexpected result as several studies reported (e.g., Similarly, when the WISSTON indicates a cold event,
Allan et al. 2001; Xie et al. 2002; Baquero-Bernal et al. it is mostly in the La Niña phase, in the tropical Pacific,
2002; Wang et al. 2003). That is, warming and cooling and then the El Niño rapidly develops during the next
events over the western Indian Ocean are related to El year, as listed in Table 2. Though some exceptional
Niño and La Niña events. It is worthwhile to note that cases exist, such as the years 1960 and 1974, Tables 1
not all ENSO events concur with the Indian Ocean SST and 2 show clear evidence for the close relation be-
events. For example, the 1965/66, 1968/69, 1976/77, tween ENSO and the Indian Ocean.
1986/87, and 1991/92 El Niño events were not accom-
panied by Indian Ocean warming.
Another interesting relation is found in Fig. 2. Some TABLE 1. The warm WISST events and the ENSO phases of the
El Niño events concurred with the Indian Ocean warm- corresponding year and following year [November–January
(NDJ)].
ing as we mentioned. In addition, in the following year,
the tropical Pacific is mostly in its cold phase. To clearly ENSO Next year
show this, we have summarized these relations in Positive during ENSO during
Tables 1 and 2. There are five cases where WISSTON is WISST yr ND(0)J(1) ND(1)J(2)
larger than its standard deviation as listed in Table 1. 1963 Weak El Niño Weak La Niña
1972 El Niño La Niña
1982 El Niño Weak La Niña
1 1987 El Niño La Niña
The subscript “ON” denotes the averaged value for October–
1997 El Niño La Niña
November.1788 JOURNAL OF CLIMATE VOLUME 19
TABLE 2. Same as in Table 1, except for cold WISST events. al. 2001). It is worthwhile to note that WISST is nega-
tively correlated to the eastern Pacific SST during the
ENSO Next-year
Negative during ENSO during following winter with a 95% confidence level. Roughly
WISST yr ND(0)J(1) ND(1)J(2) speaking, it is a 13-month lag relation. These results are
1960 Normal Normal
shown in Tables 1 and 2.
1964 Weak La Niña El Niño So far, we showed that WISSTON has a close rela-
1971 La Niña El Niño tionship with a fast ENSO transition. However, one
1974 Weak La Niña La Niña may doubt that the relationship is just a consequence
1975 La Niña Weak El Niño introduced by internal ENSO dynamics since WISST is
1985 Normal El Niño
1996 Normal El Niño
well correlated with ENSO variation. To clarify this
problem, we removed the partial influence of ENSO on
the wind and SST anomalies. To do this, we introduce
the partial correlation (Cohen and Cohen 1983). Saji
Figure 3 shows lag correlations between WISSTON and Yamagata (2003) used this method in order to
and the equatorial SST (5°S–5°N) during the years show characteristics of the Indian Ocean dipole mode,
1958–2000. Over the Indian Ocean, the correlation is which is not introduced by ENSO.
high owing to their autocorrelation. The autocorrela- Figure 4 shows the partial correlation coefficients of
tions of WISSTON are significant from the previous the equatorial zonal wind (5°S–5°N) on Niño-3.4 SST
summer to the next spring with a 95% confidence level. during November–January (dashed line) and WISSTON
In the central to eastern Pacific, the correlations are (solid line), after accounting for the effect of the other,
also significant with a 99% confidence level. It is con- respectively. When the effect of WISSTON is removed,
sistent with the results in Fig. 2 and Tables 1 and 2. the warm Niño-3.4 SST is well correlated with wester-
Note that the correlation in the central Pacific is rela- lies in the central Pacific and the easterlies in the Indian
tively high when tropical Pacific SST does not lag but Ocean. However, in the western Pacific, there are op-
leads WISSTON. In addition, WISSTON is negatively posite signs between east and west of 130°E, inferring
correlated with SST anomalies over the eastern Indian weak contribution of the Niño-3.4 SST to the western
and western Pacific Oceans with a 95% confidence Pacific zonal wind anomaly. On the other hand,
level. One may imagine that the negative correlation WISSTON has a strong correlation with the western Pa-
over the eastern Indian Ocean is related to the so-called cific wind when the effect of the Niño-3.4 SST is ex-
Indian Ocean dipole mode2 (Saji et al. 1999). In fact, cluded. In addition, there is a negative correlation in
the correlation between WISST and the Indian dipole the central Pacific, inferring that the Indian Ocean SST
index from Saji et al. (1999) is 0.76. Most WISST events anomaly plays a role of not only enhancing the anoma-
from Tables 1 and 2 were coincident with the Indian lous easterlies in the western Pacific but also partly
Ocean dipole events although there are some excep- reducing the ENSO-induced anomalous westerlies in
tional cases. For example, the 1987 warm Indian Ocean the central Pacific.
case and the 1985 cold Indian Ocean case were not Figure 5 shows the relative contributions of Niño-3.4
coincided with the positive and negative dipole events, SST during November–January and WISSTON on the
respectively. evolution of equatorial SST (5°S–5°N), representing
In addition, WISSTON is well correlated with SST the partial correlation coefficient. The results clearly
anomalies over the eastern Indian and western Pacific show that the Niño-3.4 SST is related to the western
Oceans during the following summer as shown in Fig. 3. Pacific SST cooling during the mature phase and to the
Once the warm (cold) SST anomalies are established, Indian Ocean warming during the following spring
they can play a role of expediting the development of (Klein et al. 1999; Xie et al. 2002). On the other hand,
La Niña (El Niño) by generating and maintaining WISSTON is responsible for the western Pacific warm-
anomalous easterlies (westerlies) as previous studies ing and the eastern Pacific cooling during the follow-
pointed out (e.g., Weisberg and Wang 1997a; Wang et ing year. These results support our argument that
WISSTON leads to a fast transition to La Niña.
So far, we have shown a clear covariability of the
2
There is a vigorous debate within the scientific community tropical Pacific and Indian Oceans. These results can be
concerning the physical reality of the dipole mode in the Indian
interpreted as the following processes. During a sum-
Ocean and its terminology (e.g., Hastenrath 2002, 2003; Yamagata
et al. 2003; Lau and Nath 2004). Nevertheless, in this study we just mer in which El Niño is developing, the warm SST
use the terminology of “Indian Ocean dipole mode” for the sake anomaly in the tropical Pacific modulates the Walker
of simplicity. circulation by enhancing convection in the central Pa-1 MAY 2006 KUG AND KANG 1789
FIG. 3. Same as in Fig. 1, except for WISSTON.
cific and suppressing convection in the Maritime Con- Once the warm SST anomalies are established, it can
tinent. As a result, anomalous easterly wind stress is switch to a local air–sea feedback over the Indian
induced over the Indian Ocean. The wind stress and the Ocean and this then enhances the easterly wind
related wind stress curl produce downwelling Rossby anomaly in the equatorial Indian Ocean (Webster et al.
waves in off-equator regions and lead to SST warming 1999). Therefore, a warm SST anomaly further devel-
in the western side of the Indian Ocean (Webster et al. ops in the western Indian Ocean associated with
1999; Li et al. 2002; Xie et al. 2002). In addition, a weak ENSO. To a large extent, these processes are very simi-
Indian summer monsoon, which negatively correlated lar to the developing mechanism of the Indian Ocean
to the eastern Pacific SST (Shukla and Paolino 1983; dipole mode (Saji et al. 1999; Webster et al. 1999), al-
Webster and Palmer 1997), tends to reduce the cross- though external ENSO forcing is necessary for trigger-
equatorial wind over the western Indian Ocean, infer- ing the air–sea coupled feedback over the Indian Ocean
ring that the SST is warming by reducing latent heat (Allan et al. 2001; Lau and Nath 2004).
flux. As previously mentioned, the warm SST in the west-
FIG. 4. The partial correlation of surface zonal wind averaged over 5°S–5°N on WISSTON
(solid line), after accounting for the effect of Niño-3.4 SST during November–January. The
dashed line is the same but for the partial correlation on Niño-3.4 SST during November–
January after accounting for the effect of WISSTON.1790 JOURNAL OF CLIMATE VOLUME 19
FIG. 5. (a) The partial correlation of the equatorial SST (5°S–5°N) on the Niño-3.4 SST during
November–January, after accounting for the effect of WISSTON. (b) Same as in (a), but for the partial
correlation on WISSTON after accounting for the effect of the Niño-3.4 SST during November–January.
The y axis denotes the calendar months of the ENSO developing year (year 0) and decaying year
(year 1).
ern Indian Ocean induces an anomalous local Walker ing of WISST. Following the mature phase, WISST
circulation accompanying the easterly wind stress over slowly decays while the Niño-3.4 SST rapidly decays
the Indian Ocean. Associated with the anomalous and represents the La Niña state in the latter part of the
Walker circulation, the easterly wind is established over second year. For the El Niño–only case, however, the
the western Pacific during the mature phase of ENSO. WISST is nearly zero, which means that there is no
As a result, the Indian Ocean SST affects ENSO’s evo- relation with the Niño-3.4 SST. Two El Niños slowly
lution by modulating the Walker circulation. decay and three El Niños develop another warm peak
To examine the above processes, we carried out a following the first warm peak. It is interesting that the
composite analysis. For the composite, five cases were El Niños of the WISST case are relatively strong com-
chosen in which WISSTON is greater than its standard pared to those of the El Niño–only case. It is conceiv-
deviation (0.30°C), as listed in Table 1. To compare the able that strong SST forcing over the tropical Pacific is
WISST events, we chose five other cases, which are El required to accompany an Indian Ocean warming
Niño events but not warm WISST events. The chosen event.
events were the 1965/66, 1968/69, 1976/77, 1986/87, and Detailed SST patterns and their related wind anoma-
1991/92 cases. The 1977/78 and 1994/95 cases were ex- lies are shown in Figs. 7 and 8. During the boreal sum-
cluded because WISSTON was relatively large though mer time of the developing year, warm SST anomalies
its value was less than the standard deviation. Hereafter develop in the central-eastern Pacific and western In-
we call the former the “WISST” composite and the dian Ocean, in the case of the WISST composite. These
latter is the “El Niño only” composite. The composite warm anomalies are related to anomalous westerlies in
analysis was carried out for 2 yr, of the developing (year the western-central Pacific and anomalous easterlies in
0) phase and the following (year 1) phase. the eastern Indian Ocean, respectively. Note that a cy-
Figure 6 shows the evolution of the Niño-3.4 SST and clonic flow over the Philippine Sea tends to enhance the
WISST index for WISST and El Niño–only composites. western Pacific westerlies.
For the WISST case, both the SST anomalies slowly During the boreal autumn time, the SST and wind
develop during the first year of the composite. It tends anomalies have been further developed by the SST and
to be that the warming of Niño-3.4 SST leads the warm- surface wind feedback (e.g., Bjerknes 1969) in the Pa-1 MAY 2006 KUG AND KANG 1791
FIG. 6. Time series of Niño-3.4 SST (solid line) and WISST (dashed line) during the El Niño
developing (year 0) and decaying year (year 1). Thick lines indicate composite mean for each
case. (a) The WISST and (b) El Niño–only cases, respectively.
cific and Indian Oceans, respectively. Also, the wester- ern Pacific. Recently, Watanabe and Jin (2002) showed,
lies in the Pacific have gradually propagated southeast- using their moist baroclinic model, that Indian Ocean
ward as Harrison (1987) described. This is accompanied warming is critical for developing the Philippine Sea
by an eastward movement of the cyclonic flow over the anticyclone as well as the local SST forcing does. The
Philippine Sea. In addition, an anticyclonic flow is es- Philippine Sea anticyclone can be directly connected to
tablished in the west of the cyclonic flow as Wang and the establishment of the equatorial wind anomalies as
Zhang (2002) mentioned. Also, an anticyclonic flow in shown in Fig. 8c. In addition, it seems that the western
the southern Indian Ocean (SIO) has been developing Pacific wind anomalies are closely related to the Indian
explosively (Wang et al. 2003). This anticyclonic flow is Ocean wind anomalies including the SIO anticyclone.
closely related to the equatorial easterlies and the Annamalai et al. (2005) also showed using their atmo-
WISST warming. spheric GCM experiments that the warm Indian Ocean
During the boreal wintertime, the wind system has SST anomalies generate an atmospheric Kelvin wave
moved eastward overall compared to previous seasons. associated with surface easterly flow over the equato-
And the wind system over the warm pool region seems rial western-central Pacific.
to be controlled by two great anticyclonic flows. While After the anomalous easterlies are established, the
the SIO anticyclone has weakened, the Philippine Sea SST anomalies in the eastern Pacific rapidly decay and
anticyclone has dramatically developed. The warm SST fall to near zero, while the western Pacific easterlies
has slightly shifted to east in the Indian Ocean, as have enhanced (Figs. 7d and 8d). And during the fol-
shown in Figs. 3 and 7. Possibly related to this evolu- lowing summer time the SST anomalies have developed
tion, anomalous easterlies are established in the west- to a La Niña state, the easterlies are further developed1792 JOURNAL OF CLIMATE VOLUME 19 FIG. 7. Composites of SST anomalies for (left) the WISST and (right) the El Niño–only cases. Light, medium, and dark shading represent the 90%, 95%, and 99% confidence levels, respectively. Year 0 and year 1 denote the year where an El Niño develops and the following year, respectively. in the central-western Pacific by the positive air–sea wind anomalies over the Indian Ocean. Also, the wind coupled feedback (Bjerknes 1969). anomalies in the western Pacific are relatively weak Unlike the WISST composite, the El Niño–only com- during the boreal summer and fall in the El Niño– posite does not show any significant SST and surface developing year (Figs. 8a,b). During the boreal winter-
1 MAY 2006 KUG AND KANG 1793
FIG. 8. Same as in Fig. 7, except for the 925-hPa wind anomalies. Thick arrows indicate significant wind vectors with 90%
confidence level.
time, the wind pattern shows anomalous westerlies in tion, we found that anomalous easterlies are in the
the equatorial Indian Ocean and anomalous northeast- equatorial western Pacific, though they are of little sig-
erlies in the northwestern Paciifc, similar to that of the nificance and their magnitude is weak. There are two
WISST composite but the magnitude is weak. In addi- possible explanations for the weak magnitude. One is1794 JOURNAL OF CLIMATE VOLUME 19 that the Indian Ocean does not work as we argued in During the boreal spring of the following year, a cold this study. The other is that the El Niños themselves are subsurface temperature anomaly has developed over weak compared to those of the WISST composite as the whole Pacific basin even though the SST represents shown in Figs. 6 and 7. To separate the two effects, we a warm phase. This anomalous cold ocean temperature calculated the ratios of variables between the WISST leads to a La Niña state. Note that the temperature and El Niño–only composites during November– anomaly is statistically significant, having a 99% confi- January. The ratio for Niño-3.4 SST is about 1.52. We dence level in the central Pacific. Once the cold SST also calculated this ratio for the central-eastern Pacific appears in the central and eastern Pacific, it is further zonal wind anomalies, averaged over 5°S–5°N, 170°E– developed by the air–sea coupled process. On the other 100°W, where anomalous westerlies dominate. We hand, for the El Niño–only composite, the subsurface found the ratio to be 1.34. The value is comparable with temperature anomaly is relatively weak in the western that of the ratio for the Niño-3.4, indicating that the Pacific since the anomalous easterly wind is weak, as wind anomalies depend on the strength of the El Niño shown in Fig. 8c. Therefore, the El Niño slowly decays strength. Next, we calculated this ratio for the WP zonal since its mature phase. wind index, averaged over 5°S–5°N, 120°–150°E. This As shown in Fig. 10, the western Pacific wind anoma- ratio is about 7.40, clearly indicating that the Indian lies significantly affect the structure of the ocean tem- Ocean SST plays a very critical role in developing the perature over whole Pacific basin, though wind forcing wind anomalies in the western Pacific. is confined to the western Pacific during an ENSO ma- Figure 9 shows anomalous Walker circulations for ture phase. The western Pacific wind anomalies excite WISST and El Niño–only composites. The zonal wind oceanic Kelvin waves, which rapidly propagate east- and pressure velocity are averaged over 5°S–5°N. A ward, then change the oceanic vertical structure. clear difference is found between the two composites during an El Niño–developing year. The WISST com- 4. Atmospheric GCM experiments posite shows two distinctive vertical-longitude cells So far, we have suggested, using the observational over the Pacific and Indian Oceans, while the El Niño– data, that the Indian Ocean SST warming affects the only composite shows a single cell over the Pacific ba- development of the anomalous easterlies in the western sin. Note that the two cells of the WISST composite Pacific and so finally influences the ENSO transition. seem to move slowly eastward as El Niño develops. Although the statistical analysis clearly shows the rela- Therefore, the anomalous surface easterlies are con- tionship between the Indian Ocean and the wind fined to the Indian Ocean during the boreal summer anomalies, additional experimental modeling may help and fall, but during winter the easterlies are extended the understanding of our arguments in the present into the western Pacific so that the Indian Ocean can study. Also, experimental modeling may explicitly affect the El Niño transition. In the case of the El Niño– separate the impact of the Indian Ocean SST from that only composite, a distinctive descending motion is of the Pacific SST forcing. found over the western Pacific, accompanied by a mas- To reveal the influence of the Indian Ocean SST on sive ascending motion associated with the warm SST in the western Pacific wind anomaly, atmospheric GCM the central-eastern Pacific. Unlike the WISST compos- experiments were designed, as listed in Table 3. The ite, however, the descending motion does not accom- GCM utilized is a GCM developed at Seoul National pany the ascending motion over the Indian Ocean dur- University (SNU). All experimental conditions are ing the boreal fall and wintertime. Therefore, the cou- identical except for SST boundary conditions. The SST pling between ENSO and the Indian Ocean does not boundary condition is obtained by superposing clima- operate. tological SSTs on seasonally varying SST anomalies. Figure 10 shows a structure of ocean temperature The SST anomalies are obtained from the composites along the equator. For the WISST composite, the sub- of El Niño and warm WISST events, as shown in Table surface temperature is overwhelmingly established in 1 (1972/73, 1982/83, 1987/88, and 1997/98). The 1963/64 the eastern Pacific during the developing phase of an El case was excluded due to its small amplitude. Never- Niño. However, a dramatic change in the ocean tem- theless, the SST patterns are very similar to those in perature occurs during the El Niño mature phase. The Fig. 7. In the EXPglo runs, SST anomalies are imposed warm temperature is corrupted in the eastern Pacific, in the global ocean. However, in the EXPpac and while cold temperatures develop explosively in the EXPind, the SST anomalies are only used over the western Pacific. The abrupt establishment of the tropical Pacific Ocean (30°S–30°N, 120°E–80°W) and anomalous easterlies in the western Pacific will be re- Indian Ocean (30°S–30°N, 40°–120°E) domains, respec- sponsible for the changes of the ocean temperature. tively. These sets of the experiments are expected to
1 MAY 2006 KUG AND KANG 1795 FIG. 9. Same as in Fig. 7, except for the streamline of the zonal wind and pressure velocity anomalies averaged over 5°S–5°N. reveal the individual and combined effects of SST 10 ensemble members were integrated. Our experimen- anomalies on the western Pacific wind variability. The tal design is very similar to that of Annamalai et al. model was integrated for 2 yr with the temporally vary- (2005), who tried to access the relative importance of ing SST anomalies in order to simulate consecutive at- remote (Pacific) versus local (Indian Ocean) SST mospheric responses from developing stage to decaying anomalies in determining the equatorial Indian Ocean stage of the El Niño. Using different initial conditions, variability.
1796 JOURNAL OF CLIMATE VOLUME 19
FIG. 10. Same as in Fig. 7, except for the ocean temperature anomalies along the equator.
Figure 11 shows simulated 925-hPa wind anomalies for the boundary conditions, the GCM simulates a simi-
during November–January. The anomalies were ob- lar pattern to the observed pattern (see Fig. 8), but the
tained by difference from the EXPcntl, which used sea- anomalies over the Indian Ocean are relatively weak
sonally varying climatological SSTs as the boundary (Fig. 11a). It is conceivable that this discrepancy for
conditions. When the global SST anomalies are used weak wind simulation over the Indian Ocean is linked1 MAY 2006 KUG AND KANG 1797
TABLE 3. Experimental designs with the atmospheric GCM. ered (Fig. 11b), the anomalous wind pattern is very
similar to that in Fig. 11a. The AGCM simulates the
Expt SST boundary condition
western Pacific easterlies when the SST forcing exists
EXPcntl Climatological SST only in the tropical Pacific. This is consistent with the
EXPglo Climatological SST ⫹ SSTA over the global ocean
EXPpac Climatological SST ⫹ SSTA over the Pacific Ocean
argument of Weisberg and Wang (1997a,b). However,
(120°E–80°W) compared to those of the EXPglo, the westerlies are
EXPind Climatological SST ⫹ SSTA over the Indian Ocean stronger in the central Pacific and the easterlies are
(40°–120°E) weaker in the western Pacific.
Figure 11c shows anomalous wind patterns when SST
to dry bias of the model climatology over the equatorial forcing exists only in the Indian Ocean. As expected,
Indian Ocean. However, the model well simulates the the AGCM did not simulate the anomalous westerlies
westerlies in the central-eastern Pacific and the easter- in the central Pacific. However, the easterlies were well
lies in the western Pacific. Note that the model also simulated over the western Pacific. Note that the am-
simulates the establishment of the Philippine Sea anti- plitude is larger than that of the EXPpac. This implies
cyclone during the El Niño mature phase. When the the Indian Ocean warming is a more effective contribu-
SST anomalies in the only Pacific domain are consid- tor to developing the anomalous easterlies rather than
FIG. 11. Simulated 925-hPa wind anomalies by atmospheric GCM: (a) EXPglo, (b) EXPpac, and (c) EXPind
experiment results as listed in Table 3.1798 JOURNAL OF CLIMATE VOLUME 19
FIG. 12. The WP zonal wind index simulated by atmospheric GCM. Solid, dashed, and
dotted lines indicate EXPglo, EXPpac, and EXPind results, respectively.
the warming in the eastern Pacific and the local cooling served data and atmospheric GCMs. Some El Niño
in the western Pacific. events are accompanied by warming in the western In-
To clearly show the relative contribution for the east- dian Ocean during the boreal fall and winter of the
erly anomalies, we showed the evolution of the WP developing year. We have suggested that the warming
zonal wind index, which was defined in the previous in the Indian Ocean is linked to easterly wind anoma-
section. The model simulates an abrupt establishment lies over Indonesia and the western edge of the Pacific
of the easterly anomaly in November of the El Niño– during the mature phase of El Niño. The anomalous
developing year as shown in Fig. 12. Note that the easterlies induce a fast transition to La Niña by gener-
abrupt establishment commonly occurs in both cases of ating negative Kelvin waves. Thus, the warming in the
the EXPpac and EXPind. It is conceivable that this Indian Ocean, which is a part of the El Niño signal,
feature is related to the establishment of the Philippine operates as a negative feedback mechanism to ENSO.
Sea anticyclone (Wang et al. 2000; Wang and Zhang We suggest that this Indian Ocean feedback appears to
2002). operate mostly for strong El Niños and results in a La
To compare the relative contributions of the Pacific Niña one year after the mature phase of El Niño.
and Indian Ocean SST, we calculated the WP zonal In terms of the Indian Ocean feedback, the eastward
wind index during November–January. In the case of extension of the anomalous Walker circulation is a cru-
the EXPglo, the index is ⫺1.96 m s⫺1. By contrast, the cial factor. This is linked to the establishment of the
EXPpac and EXPind simulate ⫺0.84 and ⫺1.28 m s⫺1, easterly wind anomalies in the western Pacific during
respectively. The value of the EXPind is 1.5 times an El Niño mature phase. Though the precise dynamics
larger than that of the EXPpac. Although the ratio is
for the extension of the easterly wind are unknown, we
smaller than the observational value that was presented
now suggest two possible ways. First, it seems to be
in the previous section, this result supports our argu-
related to seasonal changes of the climatological basic
ment that the Indian Ocean SST is a crucial contributor
state. There are significant seasonal changes of the sur-
to the western Pacific wind anomalies.
face wind in the Maritime Continent region (100°–
150°E). Related to these changes, the Philippine Sea
5. Summary and discussion anticyclone is dramatically established (Wang et al.
In this study, an interactive feedback between ENSO 2001; Wang and Zhang 2002), and the SIO anticyclone
and the Indian Ocean was investigated by using ob- is suddenly decayed (Wang et al. 2003). In particular,1 MAY 2006 KUG AND KANG 1799
Wang and Zhang (2002) pointed out that the develop- western Pacific thermocline and eastern Pacific SST
ment of the Philippine Sea anticyclone depends on the anomalies, respectively. Here aI and cI indicate cou-
seasonal basic state, which influences tropical– pling strength and damping, respectively. The other co-
extratropical interaction, and monsoon–ocean interac- efficients (r, ␣, b, R, and ␥) are identical to those in Jin
tion. Associated with the development of the Philippine (1997a). Though the parameter aI has a strong seasonal
Sea anticyclone and the decay of the SIO anticyclone, dependency, we ignored it for simplicity. The Indian
the anomalous Walker circulation can be expanded to Ocean SST warming induces anomalous easterly wind
the western Pacific. stress, which plays a role in producing negative ther-
The second possibility for the eastward expansion of mocline depth in the western Pacific. This process is
the anomalous Walker circulation may be related to the additionally included in the first equation of the origi-
cold anomalies in the eastern Indian Ocean. As we nal recharge oscillator model. The Indian Ocean SST is
mentioned before, most WISST events were accompa- affected by the tropical Pacific forcing, local air–sea
nied with the Indian Ocean dipole events except one interaction feedback, and local damping processes. The
case. This indicates that there are cold SST anomalies third equation represents these processes.
in the eastern Indian Ocean during these events. The When aI is zero, the equation is identical to Jin’s
warm SST in the western Indian Ocean enhances con- (1997a) model. We calculated the eigenmodes of Eq.
vection accompanying an anomalous low-level conver- (1). The value of each parameter is the same as that of
gence, and the cold SST suppresses a convection ac- Jin’s (1997a) neutral solution. As Jin (1997a) showed,
companying anomalous low-level divergence. The sup- the period of the dominant mode is about 3.4 yr3 when
pressed convection can play a role to confine the the Indian Ocean SST is decoupled from the tropical
easterly anomalies to the Indian Ocean. However, the Pacific. When aI ⫽ 1 (i.e., the ENSO is coupled to the
cold SST in the eastern Indian Ocean abruptly vanishes Indian Ocean), the period of the system is significantly
after November. Hence, the anomalous descending mo- shortened to about 2.3 yr. It means the coupling with
tions in the eastern Indian Ocean suddenly decrease. the Indian Ocean reduces the ENSO period. In addi-
Therefore, the easterly wind anomalies can be ex- tion, the eigenvector of the dominant mode shows that
tended to the western Pacific. Recently, Annamalai et the Pacific SST leads the Indian Ocean SST by 2
al. (2005) showed from atmospheric GCM experiments months. These results are consistent with the observa-
that the east–west contrast in the Indian Ocean SST tional analysis. The growth rate slightly increased com-
does not generate a significant atmospheric Kelvin pared to that of the noncoupling case.
wave response, and there is little effect in the tropical This solution of the simple analog model has some
Pacific wind variability, while the basinwide SST implication for Indian Ocean–ENSO coupling. The
anomalies can influence on the tropical Pacific wind tropical Pacific may have an intrinsic mode having a
variability. Their results somewhat support our hypoth- 3–5-yr time scale without coupling with the Indian
esis. Though we have suggested two possibilities here, Ocean. The classical ENSO theories such as the de-
further research should be devoted to precisely under- layed oscillator (Schopf and Suarez 1988; Battisti and
stand the eastward expansion of the anomalous Walker Hirst 1989), and the recharge oscillator (Jin 1997a,b),
circulation. have tried to explain the self-sustained oscillation in the
The coupled feedback process, suggested in this tropical Pacific basin. As we pointed out here, Indian
study, may be understood using a simple analog model. Ocean SST forcing modulates the ENSO mode and so
Here, we have modified the analog model for the re- changes its periodicity and growth rate (Yu et al. 2002;
charge oscillator (Jin 1997a). An equation for the In- Wu and Kirtman 2004). In particular, the Indian Ocean
dian Ocean SST is added to the original recharge oscil- SST affects the ENSO mode by producing a wind
lator model (Jin 1997a) as follows: anomaly in the western Pacific, which changes the oce-
anic thermocline.
dhw
⫽ ⫺rhw ⫺ ␣bTe ⫺ aITI Our findings, in the present study, may have useful
dt implications for ENSO prediction. A number of El
dTe Niño prediction models were developed based on the
⫽ RTe ⫹ ␥hw , 共1兲
dt tropical Pacific domain. Therefore, they cannot con-
dTI sider the western Pacific wind variability associated
⫽ bITe ⫺ cITI
dt
3
The period of the dominant mode was obtained by the recip-
where TI denotes the SST anomaly in the Indian rocal of the imaginary part of the eigenvalue, which indicates the
Ocean. As Jin (1997a) indicated, hw and Te denote the frequency of the mode.1800 JOURNAL OF CLIMATE VOLUME 19
with Indian Ocean SST. It means they may not cor- Part II: A stripped-down coupled model. J. Atmos. Sci., 54,
rectly predict a fast phase transition of ENSO. For ex- 830–847.
ample, an El Niño prediction model, developed by Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Re-
analysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.
Kang and Kug (2000), failed to predict fast-decaying
Kang, I.-S., and J.-S. Kug, 2000: An El-Niño prediction system
and -developing La Niñas following 1982/83, 1987/88, with an intermediate ocean and a statistical atmosphere
and 1997/98 El Niños though it has quite good predic- model. Geophys. Res. Lett., 27, 1167–1170.
tive skill (Kang and Kug 2000; Kug et al. 2001). These Kim, J.-K., I.-S. Kang, and C.-H. Ho, 1998: East Asian summer
deficiencies will be overcome by considering the impact monsoon simulated by the Seoul National University GCM.
of the Indian Ocean SST. This kind of approach can be Proc. Int. Conf. on Monsoon and Hydrologic Cycle, Kyongju,
studied further in order to improve ENSO prediction Korea, Korean Meteorological Society, 66–69.
Klein, S. A., B. J. Soden, and N. G. Lau, 1999: Remote sea surface
skill.
temperature variation during ENSO: Evidence for a tropical
atmosphere bridge. J. Climate, 12, 917–932.
Acknowledgments. The authors thank Drs. Bin Wang Krishnamurthy, V., and B. P. Kirtman, 2003: Variability of the
and Fei-Fei Jin for their invaluable comments and dis- Indian Ocean: Relation to monsoon and ENSO. Quart. J.
cussions. Helpful comments by anonymous reviewers Roy. Meteor. Soc., 129, 1623–1646.
have significantly improved the manuscript. This re- Kug, J.-S., I.-S. Kang, and S. E. Zebiak, 2001: The impacts of the
search was partly supported by the SRC Program of the model assimilated wind stress data in the initialization of an
Korean Science and Engineering Foundation. I.-S. intermediate ocean and the ENSO predictability. Geophys.
Res. Lett., 28, 3713, doi:10.1029/2000GL012793.
Kang was supported by Ministry of Environment as
——, S.-I. An, F.-F. Jin, and I.-S. Kang, 2005: Preconditions for El
“The Eco-technopia 21 Project.” Niño and La Niña onsets and their relation to the Indian
Ocean. Geophys. Res. Lett., 32, L05706, doi:10.1029/
REFERENCES 2004GL021674.
Lau, N.-C., and M. J. Nath, 2000: Impact of ENSO on the vari-
Allan, R., and Coauthors, 2001: Is there an Indian Ocean dipole, ability of the Asian–Australian monsoons as simulated in
and is it independent of the El Niño–Southern Oscillation? GCM experiments. J. Climate, 13, 4287–4309.
CLIVAR Exchanges, Vol. 6, No. 3, 18–22.
——, and ——, 2003: Atmosphere–ocean variations in the Indo–
An, S.-I., and I.-S. Kang, 2001: Tropical Pacific basin-wide adjust-
Pacific sector during ENSO episodes. J. Climate, 16, 3–20.
ment and oceanic waves. Geophys. Res. Lett., 28, 3975–3978.
Annamalai, H., S.-P. Xie, J. P. McCreary, and R. Murtugudde, ——, and ——, 2004: Simulation of tropical Indian Ocean SST
2005: Impact of Indian Ocean sea surface temperature on variability with east–west asymmetry using a coupled atmo-
developing El Niño. J. Climate, 18, 302–319. sphere–ocean GCM. J. Climate, 17, 245–265.
Baquero-Bernal, A., M. Latif, and L. Stephanie, 2002: On dipole- Li, T., Y. Zhang, E. Lu, and D. Wang, 2002: Relative role of
like variability of sea surface temperature in the tropical In- dynamic and thermodynamic processes in the development
dian Ocean. J. Climate, 15, 1358–1368. of the Indian Ocean dipole: An OGCM diagnosis. Geophys.
Battisti, D., and A. C. Hirst, 1989: Interannual variability in a Res. Lett., 29, 2110, doi:10.1029/2002GL015789.
tropical atmosphere-ocean model: Influence of the basic Misra, V., 2004: The teleconnection between the western Indian
state, ocean geometry, and nonlinearity. J. Atmos. Sci., 46, and the western Pacific Oceans. Mon. Wea. Rev., 132, 445–
1687–1712. 455.
Bjerknes, J., 1969: Atmospheric teleconnections from the equa- Moorthi, S., and M. J. Suarez, 1992: Relaxed Arakawa–Schubert:
torial Pacific. Mon. Wea. Rev., 97, 163–172. A parameterization of moist convection for general circula-
Carton, J. A., G. Chepurin, X. Cao, and B. Giese, 2000a: A Simple tion models. Mon. Wea. Rev., 120, 978–1002.
Ocean Data Assimilation analysis of the global upper ocean Nakajima, T., M. Tsukamoto, Y. Tsushima, and A. Numaguti,
1950–95. Part I: Methodology. J. Phys. Oceanogr., 30, 294– 1995: Modelling of the radiative process in a AGCM. Climate
309. System Dynamics and Modelling, Vol. I-3, T. Matsuno, Ed.,
——, ——, and ——, 2000b: A Simple Ocean Data Assimilation Center for Climate System Research, 104–123.
analysis of the global upper ocean 1950–95. Part II: Results.
Reynolds, R. W., 1988: A real-time global sea surface temperature
J. Phys. Oceanogr., 30, 311–326.
analysis. J. Climate, 1, 75–86.
Cohen, J., and P. Cohen, 1983: Applied Multi Regression/
Correlation Analysis for the Behavioral Sciences. Lawrence ——, and T. M. Smith, 1994: Improved global sea surface tem-
Erlbam Associate, 545 pp. perature analyses using optimum interpolation. J. Climate, 7,
Harrison, D. E., 1987: Monthly mean island surface winds in the 929–948.
central tropical Pacific and El Niño. Mon. Wea. Rev., 115, Saji, N. H., and T. Yamagata, 2003: Structure of SST and surface
3133–3145. wind variability during Indian Ocean dipole mode events:
Hastenrath, S., 2002: Dipoles, temperature gradients, and tropical COADS observations. J. Climate, 16, 2735–2751.
climate anomalies. Bull. Amer. Meteor. Soc., 83, 735–740. ——, B. N. Goswami, P. N. Vinayachandran, and T. Yamagata,
——, 2003: Reply. Bull. Amer. Meteor. Soc., 84, 1424–1425. 1999: A dipole mode in the tropical Indian Ocean. Nature,
Jin, F.-F., 1997a: An equatorial ocean recharge paradigm for 401, 360–363.
ENSO. Part I: Conceptual model. J. Atmos. Sci., 54, 811–829. Schopf, P. S., and M. J. Suarez, 1988: Vacillations in a coupled
——, 1997b: An equatorial ocean recharge paradigm for ENSO. ocean-atmosphere model. J. Atmos. Sci., 45, 549–566.1 MAY 2006 KUG AND KANG 1801
Shukla, J., and D. A. Paolino, 1983: The Southern Oscillation and ENSO. Geophys. Res. Lett., 29, 1478, doi:10.1029/
long range forecasting of the summer monsoon anomalies 2001GL014318.
over India. Mon. Wea. Rev., 111, 1830–1837. ——, and ——, 2003: A moist linear baroclinic model: Coupled
Venzke, S., M. Latif, and A. Villwock, 2000: The coupled GCM dynamical–convective response to El Niño. J. Climate, 16,
ECHO-2. Part II: Indian Ocean response to ENSO. J. Cli- 1121–1140.
mate, 13, 1371–1383. Webster, P. J., and T. N. Palmer, 1997: The past and the future of
Wang, B., and Q. Zhang, 2002: Pacific–East Asian teleconnection. El Niño. Nature, 390, 562–564.
Part II: How the Philippine Sea anticyclone established dur- ——, A. M. Moore, J. P. Loschnigg, and R. R. Leben, 1999:
ing the development of El Niño. J. Climate, 15, 3252–3265. Coupled ocean-atmosphere dynamics in the Indian Ocean
——, R. Wu, and R. Lukas, 1999: Roles of the western North during 1997-98. Nature, 401, 356–360.
Pacific wind variation in thermocline adjustment and ENSO Weisberg, R. H., and C. Wang, 1997a: Slow variability in the equa-
phase transition. J. Meteor. Soc. Japan, 77, 1–16. torial west-central Pacific in relation to ENSO. J. Climate, 10,
1998–2017.
——, ——, and X. Fu, 2000: Pacific–East Asia teleconnection:
——, and ——, 1997b: A western Pacific oscillator paradigm for
How does ENSO affect East Asian climate? J. Climate, 13,
the El Niño-Southern Oscillation. Geophys. Res. Lett., 24,
1517–1536.
779–782.
——, ——, R. Lukas, and S. I. An, 2001: A possible mechanism Wu, R., and B. Kirtman, 2004: Understanding the impacts of the
for ENSO turnabout. Dynamics of Atmospheric General cir- Indian Ocean on ENSO variability in a coupled GCM. J.
culation and Climate, IAP/Academia Sinica, Ed., China Me- Climate, 17, 4019–4031.
teorological Press, 552–578. Xie, S.-P., H. Annamalai, F. A. Schott, and J. P. McCreary Jr.,
——, ——, and T. Li, 2003: Atmosphere–warm ocean interaction 2002: Structure and mechanisms of South Indian Ocean cli-
and its impacts on Asian–Australian monsoon variation. J. mate variability. J. Climate, 15, 864–878.
Climate, 16, 1195–1211. Yamagata, T., S. K. Behera, S. A. Rao, and H. N. Saji, 2003: Com-
Wang, C., R. H. Weisberg, and J. I. Virmani, 1999: Western Pa- ments on “Dipoles, temperature gradients, and tropical cli-
cific interannual variability associated with the El Niño- mate anomalies.” Bull. Amer. Meteor. Soc., 84, 1418–1422.
Southern Oscillation. J. Geophys. Res., 104, 5131–5149. Yu, J.-Y., C. R. Mechoso, J. C. McWilliams, and A. Arakawa,
Watanabe, M., and F.-F. Jin, 2002: Role of Indian Ocean warming 2002: Impacts of the Indian Ocean on the ENSO cycle. Geo-
in the development of the Philippine Sea anticyclone during phys. Res. Lett., 29, 1204, doi:10.1029/2001GL014098.You can also read
