You are on page 1of 14

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, D14113, doi:10.

1029/2009JD013389, 2010
Click
Here
for
Full
Article

Characteristics and origin of quasi‐biweekly oscillation


over the western North Pacific during boreal summer
Guanghua Chen1 and Chung‐Hsiung Sui2
Received 16 October 2009; revised 22 February 2010; accepted 5 March 2010; published 27 July 2010.
[1] This study investigates the structure, energetic, and origin of quasi‐biweekly
oscillation (QBWO) over the western North Pacific (WNP), using NCEP reanalyses for the
years 2000–2007. In the context of vorticity there appears to be a significant QBWO mode
over the WNP during the summer. QBWO emerges from the equatorial region and
propagates northwestward. Its horizontal structure exhibits a slight southwest‐northeast tilt
but mainly longitudinal elongation. In the vertical the QBWO has a northwest tilt with
height that gives rise to a structure of the first baroclinic mode. The centers of vorticity and
vertical motion near the equator show a phase lag of about one‐quarter wavelength,
consistent with the characteristics of equatorial waves, whereas the cyclonic circulation is
tightly coupled with anomalous convection as the wave moves away from the equator.
Energetic analysis of the QBWO reveals that diabatic heating in the tropics and baroclinic
processes in the subtropics play important roles in the generation of eddy available
potential energy (EAPE). In turn, the conversion from EAPE to eddy kinetic energy (EKE)
and the barotropic conversion are major sources for EKE to compensate the loss by EKE
redistribution and dissipation. Tracing the QBWO to equatorial disturbances, our results
show some features of equatorially trapped n = 1 Rossby mode, such as phase speed and
group velocity. This mode is generally characterized by a zonal planetary wave number of
about 6 and nearly symmetric circulation about the equator. A typical case from 2002
is chosen to illustrate that the origin of the QBWO is closely associated with the theoretical
equatorial Rossby wave.
Citation: Chen, G., and C.‐H. Sui (2010), Characteristics and origin of quasi‐biweekly oscillation over the western North
Pacific during boreal summer, J. Geophys. Res., 115, D14113, doi:10.1029/2009JD013389.

1. Introduction turbances during the boreal summer. Westward‐propagating


synoptic disturbances, mainly consisting of easterly waves
[2] The boreal summer intraseasonal oscillation (ISO)
[Chang, 1970; Reed and Recker, 1971] and mixed Rossby‐
plays a key role in convection and circulation related to the
gravity modes [e.g., Yanai et al., 1968; Liebmann and
summer monsoon over the western North Pacific (WNP).
Hendon, 1990; Takayabu and Nitta, 1993], can be closely
The ISO has increasingly attracted the attention of meteor-
associated with tropical cyclogenesis over the WNP. East-
ologists over the past decades. The boreal summer ISO
ward‐propagating Kelvin waves, as well as low‐frequency
contains a variety of modes characterized by both zonal and
Madden‐Julian oscillation (MJO [e.g., Madden and Julian,
meridional propagation and nonuniform periodicity in some
1994]), have a large influence on convection over the
circumstances. Studies of linkages of ISO characteristics
Pacific [Dunkerton and Baldwin, 1995; Straub and Kiladis,
with the evolution of summer monsoon and ISO‐related 2002, 2003]. Less attention has been paid to 10–20 day ISO
synoptic phenomena have advanced our understanding of
(hereafter referred to as quasi‐biweekly oscillation (QBWO)),
the tropical atmospheric variability and may enhance pre-
which is also an important mode over the tropical Pacific
dictions of intraseasonal variations. during summer [Chen and Chen, 1993; Fukutomi and
[3] A great deal of research has been carried out investi-
Yasunari, 1999, 2002; Kang et al., 1999; Ko and Hsu, 2006,
gating convectively coupled waves over the tropical Pacific
2009]. Li and Zhou [1995] found that in the tropics, the
on the time scales of synoptic and low‐frequency dis-
kinetic energy of QBWO is much stronger than that of MJO
in either boreal winter or summer by comparing the QBWO
and MJO at 200 hPa and appears to be prominent at 850 hPa
1
Center for Monsoon System Research, Institute of Atmospheric during the July–October season. They also pointed out that
Physics, Chinese Academy of Sciences, Beijing, China.
2
Institute of Hydrological and Oceanic Sciences and Department of
the structure and activity of QBWO could differ from those
Atmospheric Sciences, National Central University, Jhongli City, Taiwan. of MJO from the viewpoint of dynamic process.
[4] The activity of QBWO can exert a significant influence
Copyright 2010 by the American Geophysical Union. on the onset (break) of the South Asian and East Asian
0148‐0227/10/2009JD013389

D14113 1 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

summer monsoon and precipitation in the Asia‐Pacific ysis, daily outgoing longwave radiation (OLR) data from the
region. For example, Chen and Chen [1995] noted the National Oceanic and Atmospheric Administration (NOAA)
impact of a 12–24 day mode on the onset and break of the satellites on a 2.5° latitude‐longitude grid for the same
South China Sea monsoon in a monsoon experiment during period of interest were used as a proxy for deep tropical
1979. Lau and Yang [1996] found rainfall oscillations with a convection [Liebmann and Smith, 1996].
distinct time scale of about 15–20 days in the Indo‐China [8] Lanczos temporal filtering was performed to extract
and East Asian region. These oscillations become remark- the QBWO signal. For 10–20 day oscillation, a Lanczos
able after the sudden jump of the Intertropical Convergence digital filter with 53 daily weights can provide very sharp
Zone from the equator to 10°–20°N, which corresponds to cutoffs of response, with negligible Gibbs oscillation. Fifty‐
the onset of monsoon convection in mid‐May to early June. three is the minimum total number of weights required to
Wen and Zhang [2008] explored the QBWO of tropical achieve unit response at the band center [Duchon, 1979]. To
convection around Sumatra and its relation to low‐level highlight the QBWO the annual cycle of the series was
circulation over the tropical Indian Ocean in boreal spring. removed by subtracting the first four Fourier harmonics
[5] Though the linkage of QBWO with the Asian summer before the Lanczos filter was employed. Unlike the MJO,
monsoon has been recognized, and some common features whose signal can be represented by zonal westerlies or
of the QBWO mode in these regions have been documented, OLR, the vorticity was chosen to identify the wave train
a more generalized view of this mode based on data for associated with QBWO, considering its smaller spatial scale
many years has not been established yet. Little is known of compared to the MJO. The empirical orthogonal function
how QBWO is developed and maintained through dynamic (EOF) of filtered vorticity was performed to identify the
processes and energy conversion associated with the back- horizontal and vertical structures of QBWO.
ground mean state. Such an understanding can be accom- [9] Lag correlation and regression analysis were per-
plished through a detailed statistical analysis of the structures formed to examine the propagation features of wave trains.
and energy budget of QBWO over the WNP. Recently, by In general, a test with N − 2 degrees of freedom is used to
examining a wide variety of QBWO activity in a global determine whether a correlation is significant. However,
perspective, Kikuchi and Wang [2009] suggested that the owing to strong autocorrelation in time series, especially for
behaviors of the westward‐propagating QBWO mode can be filtered time series, the effective sample size (ESS) was
understood in terms of convectively coupled equatorial‐ rederived for a significance test of the correlation, following
Rossby (ER) waves in the presence of monsoon mean flow. the method proposed by Bretherton et al. [1999]. For a time
However, this origin of QBWO over the tropical western series of length N, the estimation of the number of inde-
Pacific has not been further delineated by a systematic pendent observations in the sample (i.e., ESS) is
comparison of observations with theoretical considerations
N
from the literature. Therefore, this study aims to document N* ¼ NP
1
;
the structure and three‐dimensional energy balance of ð1  j j=N Þx ð Þy ð Þ
QBWO over the WNP during summertime and to verify the ¼ð N 1Þ
mechanism of initiation associated with ER wave by tracking
its origin to the equator and comparing it with equatorial wave where N* is the ESS, and rx(t) and ry(t) are the autocor-
theory, based on long‐term observational data and one typical relation functions of two time series, respectively.
case in 2002.
[6] The paper is organized as follows: section 2 describes
the data and methodology. In section 3, power spectrum and
3. Spectrum and Wavelet Analysis
wavelet analysis are used to identify the spatial and temporal [10] The 10–20 day oscillation exhibits a longitude‐
characteristics of the significant QBWO periodicity. Section 4 dependent distribution over the WNP during the boreal
presents the evolution of QBWO horizontal and vertical summer. To examine the spatial features of this variability in
structure by regression. We also performed a lag correlation the tropical western Pacific, the distribution of the power
to confirm that the tropical origin of QBWO is consistent spectrum of unfiltered vorticity at 850 hPa, as a function of
with an n = 1 ER wave. A case study using data from 2002 is period and longitude, is displayed in Figure 1. The power
detailed to further support the climatological evidence. Our spectrum was first calculated for each 123 day segment
results are summarized in the final section. (JASO season) at every grid. Then the spectra were averaged
over latitudes of 7.5°–12.5°N and 8 years to form a clima-
tological average. Following Zangvil [1977], the uP(u)
2. Data and Methodology distribution is plotted in Figure 1, where u is the frequency
[7] The daily mean products from the National Centers for and P(u) is the power. This transformation takes into
Environmental Prediction/National Center for Atmospheric account the substantial degree of “redness” in most atmo-
Research (NCEP/NCAR) reanalysis project [Kalnay et al., spheric spectra, such that a red noise spectrum with P(u) =
1996] served as the main data set for this study. The data 1/u would appear as a uniform distribution. Furthermore,
set spans from July to October (JASO) for 2000–2007 and this approach also highlights the spectral characteristics of
has a horizontal resolution of 2.5° × 2.5° at 17 pressure tropical disturbances, so it is easier to detect the spectrum
levels. It has been pointed out that while winds and air extrema in Figure 1.
temperature should be close to observations, other variables [11] Figure 1 shows that the power spectrum of low‐tro-
such as vertical velocity can be strongly model dependent pospheric vorticity has two spectral peaks between 130°E
[Kistler et al., 2001]. Apart from the NCEP/NCAR reanal- and 150°E, highlighted by shading for spectral values

2 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

spectrum power than 10–30 day band waves in terms of


different variables over the WNP. The fact that a significant
spectral extremum related to QBWO exists near 10°N, 150°E
is consistent with previous studies. For example, Kiladis
and Wheeler [1995] documented that the 6–30 day filtered
voticity centers related to equatorial waves are located near
10°N (see their Figures 2 and 7a). The composite of 7–30 day
filtered 850 hPa stream function by Ko and Hsu [2006] also
showed that the submonthly circulation can be tracked back
to 150°E near the equator (see their Figure 5).
[12] To validate the spectral feature and seasonal distri-
bution for the boxed region in Figure 1, a wavelet analysis
was carried out by applying an approximate cosine taper to
the time series to reduce end effects and sidelobe leakage.
Using the same averaging procedure as applied to data
shown in Figures 1 and 2 reveals two significant bands of
wavelet spectrum, one being almost uniformly distributed
near a 6 day period and the other appearing along the 10–
20 day period band. These features are consistent with the
Figure 1. Multiyear mean power spectrum of 850 hPa spectrum distribution near 150°E in Figure 1. Here use of
vorticity multiplied by frequency, uP(u), averaged over the significance test verified that the QBWO variation is
7.5°–12.5°N as a function of period and longitude for the statistically significant from July to October.
July–October season over 8 years (2000–2007). Contour [13] In addition, the standard deviation of the 10–20 day
intervals are 10−11 s−2 day−1. Shading denotes values greater band‐pass‐filtered vorticity during the JASO season offers an
than 10−10 s−2 day−1, significant at the 90% confidence level. overview of the geographical features of QBWO. Figure 3
shows the spatial pattern of the standard deviation of fil-
tered vorticity, which is characterized by a northwest exten-
greater than 10−10 s−2 day−1, which is significant at the 90% sion of large variations from the equatorial region to
confidence level. The peak near 135°E coincides with a 6– subtropical East Asia. The location of extremum in the tropics
10 day time scale and is in good agreement with the synoptic near 150°E coincides with the spectral peaks identified in
disturbances observed by many previous studies [e.g., Lau Figure 1. The subtropical maximum is located 130°E north-
and Lau, 1990; Liebmann and Hendon, 1990; Chang et east of Taiwan. If the same spectrum and wavelet analysis are
al., 1996; Chen and Huang, 2009]. They unexceptionally performed at a variety of locations selected along the large
identified the disturbance as a train of southeast‐northwest‐ standard deviation area of 10–20 day filtered vorticity, they
oriented synoptic‐scale waves, generally referred to as tropical‐ also yield results qualitatively similar to those presented in
depression (TD)‐type disturbances. The other spectral peak Figures 1 and 2.
is located near 150°E, with a period of 10–20 days (see the
boxed area in Figure 1), suggesting that QBWO is also an
important mode of wave disturbances in the tropical western 4. Structure and Energetics of Quasi‐biweekly
Pacific from a climatological perspective. However, it is Oscillation (QBWO)
noteworthy that the magnitude of the spectral extremum is [14] To indentify further the characteristics of the 10–
evidently smaller compared to that of TD‐type disturbance, 20 day disturbances, we analyze the structure of the dis-
which is also anticipated, since some researchers [e.g., turbances. We also perform an analysis of the energetics to
Kikuchi and Wang, 2009] similarly found that synoptic‐scale identify the relevant processes responsible for the generation
disturbances with a time scale of less than 10 days have more and maintenance of the disturbances.

Figure 2. Ensemble mean wavelet power spectrum (using the Morlet wavelet) of 850 hPa vorticity
averaged over 7.5°–12.5°N and 145°–150°E, normalized by its variance. The shaded regions mark spectra
significant at >95% confidence.

3 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Figure 3. Geographical distribution of the standard deviation of the 10–20 day band‐pass‐filtered vor-
ticity at 850 hPa for July–October 2000–2007. Contour intervals are 0.5 × 10−6 s−1. Values greater than
0.5 × 10−5 s−1 are shaded.

4.1. Horizontal and Vertical Structure of QBWO flow field is characterized by a zonally elongated structure
[15] To examine the structure of QBWO over the WNP, of alternating cyclonic and anticyclonic circulation, with
the EOF analysis is applied to the 850 hPa 10–20 day fil- dominant southwesterlies extending northeastward from the
tered vorticity field. The leading EOF spatial modes can South China Sea to the subtropical region near 170°E. The
explain the large fraction of the variance. In previous studies associated band of cyclonic vorticity is well collocated with
the rotated EOF (REOF) or extended EOF (EEOF) was also anomalous convection that covers about 60° longitude. The
used to diagnose the dominant mode of synoptic TD‐type wave train off the equator has a wavelength of approxi-
variability over the WNP [e.g., Lau and Lau, 1990; Maloney mately 3500 km and originates from the tropical region
and Dickinson, 2003]. They produced the perturbation struc- where cyclonic voticity and winds are weak. In addition, a
ture and propagation behavior consistent with the counterpart pair of equatorial convection and cyclonic circulation (the
from the EOF approach. In this study because the domain of negative and positive vorticity contours in the Southern
analysis is smaller than that used in previous studies, an EOF Hemisphere stand for cyclonic and anticyclonic circulation in
analysis yields results similar to those obtained by REOF or Figure 5) is discernible near 170°E, centered near 5°N and
EEOF (not shown). Thus, a covariance matrix of the data in 5°S, although the components in the Southern Hemisphere
the domain of 0°–30°N, 120°–170°E was constructed for are weaker and spatially smaller. At lag 0 (Figure 5b) the
EOF analysis in the present study. wave train is displaced northwestward off the equator, while
[16] Figure 4 shows the first two EOF spatial modes of the the equatorial cyclonic voticity north of the equator propa-
filtered 850 hPa vorticity field, explaining 13.3% and 12.0% gates westward and strengthens near 150°E. Enhanced con-
of the variance, respectively. The remaining modes of EOF vection emerges ahead of the cyclonic center by one eighth to
do not seem to resemble any physical phenomena (not one quarter of a wavelength. The other weak cyclonic
shown). The leading two EOFs are comprised of alternating counterpart to the south of the equator still exists, about 10°
positive and negative vorticity centers nearly aligned along longitude east of the northern one. In contrast to Figure 5a, the
the axis of standard deviation of vorticity in Figure 3. These zonally elongated positive vorticity in the subtropical region
two spatial modes are in quadrature, and their corresponding contracts in horizontal scale and propagates northwestward to
principal components (PCs) are correlated at 0.79 when PC2 the south of Japan. This part of the wave train is tightly
leads PC1 by 4 days, approximately a quarter of the oscil- coupled with active convection. The structure at lag +4 days
lating period. These indicate that the leading EOFs describe (Figure 5c) resembles that in Figure 5a, but with opposite
a northwestward‐propagating signal. Therefore, an index signs, corresponding to a one‐half‐period difference in phase.
was constructed in the following manner: Index(t) = PC1(t) + [18] As implied in Figure 5, wave development is evidently
PC2(t − 4), where t is the time in days. The index is a linear asymmetric in both hemispheres, which can be attributed to
combination of PC1 and PC2 with contributions from PC2, the asymmetric distribution of large‐scale monsoon flows
reflecting that PC2 peaks an average of 4 days before PC1. with respect to the equator. The climatological background
The horizontal and vertical structure of QBWO is explored state is depicted in Figure 6. The monsoon trough within the
by linearly regressing variable fields onto this index, all region of 0–20°N and 130–150°E has a strong cyclonic
scaled by 1 standard deviation of the index. shear (−∂u/∂y > 0). Along the monsoon trough and in its
[17] The regressed fields of OLR, vorticity, and wind vicinity, the southwesterly monsoon flows meet southeast-
vectors at 850 hPa at lags of −4, 0, and +4 days are dis- erly trade winds, giving rise to a prominent convergence of
played in Figure 5. At lag −4 days (Figure 5a) the horizontal zonal wind (−∂u/∂x > 0). Consequently, a wave accumula-
tion process tends to occur, leading to the increase in wave

4 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Figure 4. The (a) first and (b) second eigenvectors obtained from an empirical orthogonal function
(EOF) analysis of 850 hPa vorticity. Contour interval is 0.02. Negative contours are dashed.

amplitude documented in previous studies [e.g., Webster [19] In addition, as depicted in Figure 6 the monsoon
and Chang, 1988; Kuo et al., 2001]. Their studies showed trough is characterized by easterly vertical shear, which
a wave accumulation process produced by ∂u/∂x < 0, in could help trap wave energy in the lower troposphere as
which waves increase in amplitude and decrease in scale, proposed by Wang and Xie [1996] and Xie and Wang
and that equatorial waves can eventually turn poleward out [1996]. Using a two‐layer model that included para-
of the equatorial waveguide as they amplify owing to meterizations for wave‐scale friction and diabatic heating,
changes in basic flow associated with the monsoon trough. they simulated the growth of ER waves in the presence of
Sobel and Bretherton [1999] reported that, in the western spatially uniform easterly vertical shear. They found that
Pacific Ocean, convergence of the wind was the primary maximum growth occurred at wavelengths near 3500 km,
component of group velocity convergence and the wave remarkably similar to those in this study, and wave energy
accumulation process could be viewed simply as the growth was trapped in the lower troposphere. When asymmetric
of a wave or a vortex in convergent background flow. This vertical shear was imposed, with easterly shear extending
was represented in a linearized vorticity equation as the well north of the equator, ER wave growth occurred only in
product of background divergence and disturbance vorticity the Northern Hemisphere. On the contrary, the westerly
[e.g., Aiyyer and Molinari, 2003]. Once a wave begins to vertical shear north of 30°N tends to confine the wave
grow, convection is likely to become active and contribute to energy in the upper troposphere and, thus, does not favor
intensification. By this reasoning, wave accumulation pro- lower‐level wave development to a certain extent.
duces the initial intensification of the waves and is a critical [20] Overall, the background flows, including the mon-
factor that allows subsequent growth by other mechanisms. soon convergence zone and easterly vertical shear, as well as

5 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

convection and, in turn, supply the abundant moisture and


heat flux for wave growth.
[21] The vertical structure of 10–20 day filtered variable
fields is also examined in the same way. These filtered
variables along the vertical cross section of the line segment
in Figure 5b are regressed onto the QBWO index at lags of
−4 and +4 days. Figures 7a and 7b show the regressed fields
of vorticity and vertical velocity, which are scaled by 1
standard deviation of the index series. The regressed fields
at lags −4 and +4 days are similar but of opposite signs,
representing approximately a half‐period phase difference.
Vorticity perturbations peak in the lower and upper tropo-
sphere and tilt northwestward with height, so that the
structure looks like the first baroclinic mode. The vertical tilt
of vorticity is also consistent with the structure of synoptic‐
scale disturbances diagnosed by Reed and Recker [1971].
The temporal evolution indicates northwestward propaga-
tion of the perturbation vorticity anomalies. The configura-
tions of vertical velocity and vorticity are distinct in the
tropical and subtropical regions. Upward (downward) motion
is located to the southeast of the positive (negative) vorticity
anomalies at about one eighth to one quarter of a wave-
length in the tropical region south of 17°N. This implies that
the waves close to the equator may be related to the classical
equatorial waves deduced by Matsuno [1966] since a family
of equatorial waves is characterized by the phase lag of cir-
culation center (vorticity) and convergence (vertical motion).
As the waves are displaced away from the tropics, the vertical
motion maxima are almost superposed above the vorticity
centers at 23°N. This suggests that the enhanced convections
are tightly coupled with the cyclonic centers, consequently
leading to a more obvious baroclinic vertical structure, with
positive vorticity in the lower troposphere and negative vor-
ticity in the upper troposphere. The vertical structure of tem-
perature anomalies (not shown), whose centers are located
at the midtropospheric level, coincides with that of vertical
velocity, that is, the warm (cold) anomalies correspond to
the anomalous ascent (descent).
4.2. Energy Budget of QBWO
[22] In the previous discussion we documented the hori-
zontal and vertical structures of the QBWO based on the
regressed circulation and OLR fields. The characteristics of
structure and the phase lags among various variable fields
are indicative of conversions of kinetic energy and available
potential energy between the background fields and eddy
Figure 5. Regressed fields of 10–20 day filtered vorticity perturbations. Therefore, an analysis of the energy budget of
(contour), wind vectors, outgoing longwave radiation (OLR; the QBWO can improve our understanding on the devel-
shaded) onto the quasi‐biweekly oscillation (QBWO) index. opment and maintenance of the QBWO over the WNP. In
Regressions are shown for lags of (a) −4 days, (b) 0 days, and this section we estimate the conversion terms in the energy
(c) +4 days. Contour interval is 10−6 s−1. Negative contours budget to better understand the forcing of the QBWO.
are dashed. Shadings from light to dark gray denote −2, −4, [23] The energy budget formulations used here follow
and −6 W m−2. Only wind arrows significant at a level of those derived by Lau and Lau [1990]. The kinetic energy
>90% are plotted. equation is an expansion of the two‐dimensional energy
budget analysis used by Maloney and Hartmann [2001],
which only accounted for barotropic energy conversion at
large‐scale heating in the monsoon trough, are primarily
850 hPa. Any variable a in our system is divided into two
confined to the Northern Hemisphere. As a result, the parts: the spatial‐dependent JASO climatological mean state
northern wave component departs from the equator and
and 10–20 day filtered QBWO perturbation at a specific
grows rapidly, while the southern one tends to weaken and
location obtained from the regression, denoted aclim and a′,
even disappears. Furthermore, the warm sea surface tem- respectively. In this analysis a denotes an average over one
perature in the WNP would also favor the development of
wave period from −7 to +7 days of regression lag, so a0 .

6 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Figure 6. Climatological mean of wind (streamline), stretching deformation of zonal wind (shaded; units
of 10−6 s−1) at 850 mb, and vertical wind shear of −10 m s−1 (dashed line), 0 m s−1 (solid line), and 10 m s−1
(dotted line) for July to October 2000–2007.

This approach somewhat differs from previous studies of equation for the complete wave cycle. The first three terms
wave energetics for which the calculations are based on one on the right‐hand side in equation (1) are most dominant,
or more cases rather than a wave climatology [e.g., Reed et while the other terms are negligibly small.
al., 1988; Maloney and Dickinson, 2003]. In addition, the [25] Figure 8a shows the generation of EKE by barotropic
wave perturbations are defined here with respect to temporal conversion, which is mainly determined by u0 v0 @u=@y and
rather than longitudinal average. This gives a more accurate u0 u0 @u=@x. Because of the southwest‐northeast phase tilt
estimate over one wave cycle than an average over longi- of QBWO (u0 v0 > 0), the total conversion of MKE to EKE
tude, where arbitrary domain boundaries would have to be primarily depends on the pattern of mean flow related to the
drawn. In the following discussion we investigate the crucial monsoon trough and the subtropical high. In Figure 6 the
contributors in the eddy kinetic energy (EKE) and eddy monsoon trough within the region of 0°–20°N and 130°–
available potential energy (EAPE) budgets. The spatial 150°E has a strong cyclonic shear (−∂u/∂y > 0) and a con-
distributions of their vertically integrated estimates and the vergence of zonal wind (∂u/∂x < 0), while the opposite
vertical distributions along the vertical cross section (shown situation occurs north of 25°N, where the basic flow asso-
in Figure 5b) are shown in Figures 8 and 9, respectively. ciated with subtropical high is anticyclonic (−∂u/∂y < 0) and
[24] The EKE balance equation is written as divergent (∂u/∂x > 0). As a result, the barotropic conversion
from MKE to EKE is positive in the tropics and negative in
@K 0 R 0
the subtropics. The corresponding vertical cross section of
¼ Vh  ðV 0  rÞVhclim  !0 T 0  r  ðV 0 F Þ  Vclim
@t P the barotropic conversion term as displayed in Figure 9a
 rK  V 0  rK þ D; ð1Þ reveals that the barotropic conversion in the tropics is con-
fined to the lower troposphere, suggesting that the genera-
where K = (u′2 + v′2)/2 is the horizontal EKE, and other tion of EKE by barotropic process plays an important role in
symbols follow conventional definition. The first term on the genesis of QBWO perturbation at 850 hPa. In contrast,
the right‐hand side of equation (1) represents barotropic large values of EKE destruction through barotropic pro-
conversion from time‐mean kinetic energy (MKE) to EKE. cesses occur at a higher level in the tropics, especially at
The second term is conversion from EAPE to EKE through 150 hPa, and throughout the troposphere in the subtropics
the negative correlation of perturbation temperature and north of 20°N, resulting in a decrease in vertically averaged
pressure velocity. These two terms are considered true barotropic conversion or even a negative contribution to
sources or sinks of EKE. The other terms describe the EKE generation. Compared to TD‐type disturbances, the
redistribution of perturbation energy, which include the magnitude of barotropic conversion in QBWO is smaller
generation/destruction of EKE by local convergence of eddy and accounts for a lower proportion of kinetic energy
geopotential flux (the third term), the advection of EKE by sources. The differences can be explained by the fact that
the climatological mean flow and the transient fluctuations TD‐type disturbances have a pronounced southwest‐north-
(the fourth and fifth terms), and the net generation or dissi- east tilt in horizontal structure and a relatively small scale
pation of EKE by frictional and other subgrid scale effects (e.g., see Figure 7 of Chang [1996]) so that they can produce a
(the last term). We carried out an analysis of the EKE considerable barotropic conversion from MKE to EKE. In

7 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Along the track of QBWO development the upward


(downward) motion is collocated with the warm (cold)
temperature anomaly throughout the whole troposphere
(Figure 9b). As a result, the conversion from EAPE to EKE
occurs throughout the troposphere, yielding the major con-
version zone in the vertically integrated field (Figure 8b).
[27] The conversion related to the third term exhibits a
distribution similar to that
 ofthe second term, but with an
0 0 0
opposite sign. Since −r · V 0 F ¼ V 0  rF  F ðr  V 0 Þ =
0
Vh  rh F þ PR !0 T 0 , in which the term PR !0 T 0 is dominant, the
0

conversion from EAPE to EKE in the middle and upper


troposphere in the second term is almost offset by local
convergence of the eddy geopotential flux in the third term
in which the EKE in the troposphere is redistributed. In
addition, advections of EKE by the mean flow and pertur-
bation flow are generally small in most of the WNP, indi-
cating that most redistribution is accomplished by
geopotential flux.
[28] From the thermodynamic equation
@T @T @T Q
þu þv  Sp ! ¼ ; ð2Þ
@t @x @y Cp

one can derive a budget equation for EAPE (defined by A =


RT′2/2SpP) by multiplying the thermodynamic equation by
RT′/SpP, to obtain

@A R 0 0 RT 0 !0 R 0
 Vh T  rh Tclim þ þ Q T0; ð3Þ
@t Sp P P Cp Sp P 1

where Sp = −Tclim@ ln@pclim ¼ RT @Tclim


Cp P  @p is the static stability,
clim

R is the gas constant, Cp is the specific heat at constant


pressure, and Q′1 is the perturbation apparent heat source,
which can be calculated as a residual from the perturbation
temperature tendency, equation (3). The three terms on the
right‐hand side of equation (3) represent EAPE generation
or destruction through the baroclinic conversion, conversion
between EAPE and EKE, and the diabatic heating effect.
[29] The term  SRp P Vh T 0 · rhTclim represents baroclinic
0

conversion from mean available potential energy (MAPE) to


EAPE conversion. It is generally known that this baroclinic
Figure 7. Vertical cross section of regressed vorticity (con- conversion is dominant in the extratropics north of 20°N
tour; units of 10−5 s−1) and vertical velocity (shaded; units of (Figure 8c), owing to the large climatological north‐south
pa s−1) onto the QBWO index for lags of (a) −4 days and temperature gradient and the positive eddy heat flux (V′hT′ > 0)
(b) +4 days, along the line shown in Figure 5b. Contour in the lower and middle troposphere. Along the vertical cross
interval is 10−6 s−1. section of the QBWO centers (Figure 9c), the destruction of
EAPE above 200 hPa arises from the vertical baroclinic
structure (V′hT′ < 0) and the eddy heat flux directed against
addition, TD‐type disturbances have a more westward‐ the mean temperature gradient.
propagating component, which ensures that synoptic waves [30] The second term in equation (3) is the destruction of
are almost uniformly embedded in the favorable monsoon EAPE through conversion to EKE, opposite to the second
circulation south of the subtropical high, providing persistent term in EKE balance, equation (1). To supply the loss of
barotropic kinetic energy transfer from mean flow. EAPE owing to conversion from EAPE to EKE, the diabatic
[26] Figure 8b shows the conversion of EAPE to EKE process plays a decisive role in the generation of EAPE.
through the rising (sinking) motion of warm (cold) air, Although calculation of the diabatic heating term using
which accounts for a major generation of EKE over the reanalysis is problematic, the warm temperature anomaly
broad WNP region. This conversion process provides more coincides with upward motion where OLR is minimum,
appreciable energy sources for QBWO perturbation than the suggesting that latent heating from convection is coupled to
barotropic conversion process (Figure 8a). The areas of major the wave disturbance to generate EAPE. Figure 8d shows
generation of EKE from EAPE conversion in Figure 8b are that EAPE generation by perturbation diabatic heating
consistent with the orientation of QBWO development. approximately balances the EAPE conversion to EKE,

8 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Figure 8. Vertically averaged distributions of (a) the eddy kinetic energy (EKE) barotropic conversion,
(b) the conversion from eddy available potential energy (EAPE) to EKE, (c) the EAPE baroclinic conver-
sion, and (d) EAPE generation by diabatic heating. Contour interval is 10−6 m2 s−3.

especially in the tropical region. Similarly, the diabatic velocity are compared with those derived from the linear
process associated with latent heating owing to rising shallow‐water solution for the observed wavelength and
motion reaches its maximum in the middle troposphere constant background zonal flow, as by Molinari et al.
(Figure 9d). The axis of maximum diabatic heating is [2007]. These calculations follow the approach of Zhang
coincident with the axis of the QBWO growth in the tropics and Webster [1989].
as shown in Figure 5b. It is evident that the main balance of [32] We used lag correlation analysis to indicate the wave
the EAPE budget for QBWO disturbances is between the characteristics based on a reference time series of 10–20 day
EAPE‐EKE conversion and the diabatic effect. Thus the filtered vorticity at the base point (10°N, 150°E). ER waves
strong loss of EAPE through conversion to EKE is mostly can also be represented by the equatorial zonal wind as well
maintained by diabatic heating over the WNP in the tropics as off‐equatorial meridional wind (e.g., see Figure 10 of
and by baroclinic conversion from MAPE to EAPE in the Wheeler et al., 2000], which would produce the same
regions north of the subtropics. Comparing Figure 8a with qualitative results. The time series are composed of 123 ×
Figure 8d, the magnitude of generation related to diabatic 8 = 984 samples for 8 years of JASO (i.e., 8 × 123 daily
heating is much greater than that related to barotropic con- fields for the JASO period each year). Figure 10 shows the
version, indicating that latent heating is a crucial energy lag correlation of the 850 hPa vorticity at each longitude
source for the maintenance of the QBWO over the WNP. point along the 10°N latitude band with the reference vor-
ticity at (10°N, 150°E), which is chosen as a representation
5. Tropical Origin of QBWO of equatorially trapped ER wave properties in the tropical
western Pacific. The abscissa represents longitude, and time
5.1. Properties of the Wave Packet progresses downward on the ordinate. The two straight lines
[31] Although the symmetry of a pair of westward‐moving (labeled “P”) are drawn through two neighboring zero‐cor-
cyclonic (or anticyclonic) centers coexisting in both hemi- relation contours to define half a wavelength by the abscissa
spheres is, to a certain degree, lost owing to the asymmetry distance and half a period by the ordinate distance in
of the background mean state (Figure 5), the preceding between. Therefore, it is clearly shown that the wave is
analysis indicates that the QBWO is connected with equa- characterized by one wavelength of 6000 km and one period
torial waves close to the equator. In addition, comparing the of 14 days, which is consistent with a typical zonal scale of
time scale of the QBWO with those of various equatorial observed QBWO in the Asian summer monsoon region [e.g.,
waves, it is hypothesized that the origin of the westward‐ Chen and Chen, 1993]. Because of the Doppler effect the
propagating QBWO may be related to n = 1 ER waves. The equation Ud = U + Ui, where Ud is the Doppler‐shifted
symbol n is referred to as the meridional mode number, velocity and U is the basic flow, is used to compute the
denoting the number of nodes in the meridional profile of intrinsic phase velocityUi. The zonal flow U in this equation
meridional velocity. The larger n is, the more nodes the will be defined as the 850 hPa climatological mean zonal
meridional velocity has in the meridional direction. To wind at 150°E, 10°N for 8 years of the JASO period, which
support this assumption, the observed phase and group gives U = 1.02 m s−1. Based on half a wavelength and lag

9 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Figure 9. Vertical cross section of (a) the EKE barotropic conversion, (b) the conversion from EAPE to
EKE, (c) the EAPE baroclinic conversion, and (d) EAPE generation by diabatic heating along the line
shown in Figure 5b. Contour interval is 2 × 10−6 m2 s−3.

time span between two successive nodes shown in Figure 10, intrinsic velocity Ui of approximately −5.98 m s−1 obtained
Ud can be estimated as −4.96 m s−1. In contrast, the theoretical from the lag correlation.
ER wave phase velocity can be computed following the [33] Similarly, by inspection of the line labeled “G” in
pffiffiffiffiffiffiffi Ui = w/k, in which w = −bk/[k + b(2n + 1)/c0], c0 =
2
formula Figure 10, the group velocity almost stays close to 0. That
gH , and k = 2p/L. H is an important physical parameter that is, while the individual crests of the wave train form to the
determines the propagation features of the ER wave, called east of the grid point, pass through it, and decay to the west,
the equivalent depth. Using the wave number‐frequency the largest amplitude of the wave is approximately fixed at
spectra method, Wheeler et al. [2000] found that the con- the base grid point. At that point the wave appears as nearly
vective variations contributing to spectral peaks lie along the a standing oscillation. Considering the presence of basic
equatorial wave dispersion curves for equivalent depths in zonal wind as well as the previously specified equivalent
the range of 12–50 m, which corresponds to tropospheric depth and wavelength, the intrinsic group speed reasonably
vertical wavelengths of 7–13 km. Owing to a vertically matches the theoretical magnitude of −2.56 m s−1. In con-
extending structure (i.e., a relatively large vertical wave- trast, deducing from the dispersion relation w = U k − b/k for
length) of the QBWO as shown in Figure 7, the equivalent a small meridional wave number, the phase and group
depth H is set to 50 m for theoretical estimate. For a velocity for nondivergent barotropic Rossby waves with a
wavelength of 6000 km the theoretical value of wave phase wavelength of 6000 km and the same zonal basic flow, just
velocity is roughly −5.45 m s−1, comparable to the observed like the common phenomenon at midlatitudes, are −19.6 and
21.6 m s−1, respectively, far from those observed. Another

10 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

water solution for the ER wave contains convergence one‐


quarter wavelength east of the low. This convergence pattern
was reflected by the negative OLR anomalies east of the low
in Figure 11c. Three days later (Figure 11d) the northern L1
amplified and formed a strong cyclonic circulation that was
longitudinally elongated. Comparatively, the southern L1
was closer to the equator than in the previous time and,
moreover, smaller in horizontal scale than the northern L1.
[37] On 13 August (Figure 11e) the northern L1 moved
northwestward to near 15°N, 147°E while continuing to
intensify. The southern L1 had become obscure. A pair of
highs (H2) appeared near 170°E, replacing the pair of lows
(L1) that first appeared at the same location around 7 August.
The 6 day period between the emergence of L1 and that of
H2 is approximately one half of a period. Subsequently, a
southeast‐northwest‐oriented train of alternating cyclonic and
anticyclonic disturbances became evident. The wave train
was characterized by a southwest‐northeast tilt in horizontal
Figure 10. Correlation between a reference time series of
structure and a northwestward propagation. The southern
10–20 day filtered 850 hPa vorticity at 10°N, 150°E and
the same vorticity field at each grid point along 10°N lati- H2 as shown in Figure 11f, located 163°E to the east of the
northern H2, weakened and approached the equator. By
tude for various lags behind the reference time series by
19 August (Figure 11g) an enhanced convection anomaly
the number of days noted on the y axis.
appeared at the equator near 175°E, resulting in another twin
cyclonic flow (L2) straddling the equator to the west of
convectively coupled equatorial wave mode with a similar convection. The twin cyclones were not aligned along the
period is the Kelvin wave; however, it features an eastward same longitude. During the following 3 days (Figure 11h)
propagation and a representation of zonal wind. Conse- the northern L2 again moved northwestward away from the
quently, there is little doubt that the observed wave packet equator, while the southern one gradually moved westward
represents ER waves rather than any possible alternative. and weakened as another high (H3) emerged to the east,
yielding a new alternating wave train. This pattern per-
5.2. A Case Study sisted for 20 days more before eventually losing its ER
[34] This section explores a typical case in August 2002, wave signatures.
which clearly exhibited the propagation and structural evo- [38] It is worthwhile to stress that, before the northern
lution of a tropical wave in the 10–20 day filtered band. wave component moved away from the equator, the
Extracted from a complete event lasting approximately observed equatorial wave structures previously depicted
1.5 months, one episode is chosen for detailed examination to were not in strict agreement with the shallow‐water solution
support the argument that the properties of QBWO origin in derived from the assumption of a barotropic system without
the tropics can be associated with a convectively coupled n = mean flow. For example, the twin vortices did not occur at
1 ER wave in many aspects. the same longitude and were not symmetric about the
[35] Figures 11a–11h show the evolution of the wave equator. In addition, the southern component of the wave
packet in which the respective low and high twin vortices was weaker or smaller. To some degree, the above asym-
straddling the equator achieved their highest amplitude. metric features can be explained by theoretical or numerical
These weather maps are shown in 3 day increments from wave solutions in realistic three‐dimensioned basic flows.
1 August to 22 August 2002. Figure 11a does not give a Furthermore, the structures shown in Figure 11 closely
clear indication of the shallow‐water ER structure. Rather, a resemble the composite western Pacific ER wave described
high‐pressure area (indicated by “H1”), characterized by an by Wheeler et al. [2000, their Figure 9]. Their composite ER
anticyclonic circulation center at 5°N, 170°E, extended wave also contained a Southern Hemisphere low that was
longitudinally from 150°E to 180°. Along the same longi- smaller and farther east than that to the north and a nearly
tudinal band the counterpart in the Southern Hemisphere identical OLR anomaly structure. Frank and Roundy [2006]
showed a stronger anticyclonic circulation without a closed showed an ER wave composite during cyclogenesis events
center. Three days later (Figure 11b) the northern high (H1) over the WNP that is also similar to Figures 11b and 11c.
moved westward and intensified, while the southern high On the basis of the preceding analysis, although the
developed into an anticyclonic center (also labeled H1) at 7°S, observed structures differed somewhat from the shallow‐
160°E, reflecting a pair of highs like an ER wave structure. water solution, the packet is almost certainly made up of ER
At the same time, to the east of the pair of highs, a pair of waves.
symmetric convections about the equator emerged near the
dateline, indicative of the commencement of another pair of
lows. 6. Summary
[36] By 7 August (Figure 11c), H1 moved northwestward [39] This study presents a statistical analysis of the
to 15°N, 145°E, while the counterpart in the Southern structure, energetics, and origin of the QBWO over the
Hemisphere disappeared. A pair of lows originating from the tropical western Pacific based on the 8 year satellite obser-
dateline (labeled L1) formed near 165°E. The linear shallow‐ vation and reanalysis data. The QBWO signal can be iso-

11 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Figure 11. The 10–20‐day filtered 850 hPa wind field (vector) and OLR anomaly (shaded) on (a) 1 August,
(b) 4 August, (c) 7 August, (d) 10 August, (e) 13 August, (f) 16 August, (g) 19 August, and (h) 22 August
2002. Shaded areas indicate OLR anomalies of −10, −20, and −30 W m−2.

12 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

lated from observed vorticity using a 10–20 day cutoff by [43] In addition to the presence of 1 maximum standard
Lanczos filter. Since the power spectrum and wavelet of deviation at 150°E in the tropics, another extremum, with a
vorticity identify the 10–20 day period as a significant higher amplitude and a larger extent than that in the tropics,
spectral band in the tropical western Pacific during summer was found north of 20°N in East Asia (shown in Figure 3),
other than TD‐type disturbance, a combination of the two which may be partly attributable to disturbances originating
leading PCs of vorticity EOF decomposition was defined as from the midlatitude zone. Using the OLR in the South
a reference time series for regression. The similarity China Sea as an index, Fukutomi and Yasunari [1999, 2002]
between the spatial distribution of EOF of filtered vorticity found a two‐way tropical‐extratropical interaction on a 10–
and its standard deviation further reveals the signature of the 25 day time scale over the western Pacific during the early
QBWO. The wave structure is then determined by regres- summer. They examined the relationships between tropical
sing key variables like OLR, wind, and vertical velocity convection over the South China Sea and large‐scale cir-
onto the defined time series. culation and showed a lower and upper tropospheric wave
[40] The QBWO can be traced to westward‐moving train with a zonal wave number of 5–6 arcing into the
equatorial wave disturbances that reach 150°E and then North Pacific. In the lower troposphere the alternating
propagate northwestward toward south of Japan. The re- subtropical circulation anomalies propagate southwestward
gressed structure illustrates an alternating cyclonic and into the South China Sea region, following the westerly duct
anticyclonic wave train with a southeast‐northwest orienta- extending from the tropical monsoon region to the North
tion and a wavelength of about 3500 km. Its horizontal Pacific along the northern periphery of the subtropical high.
structure tilts slightly southwest‐northeastward and has a Considering their results, it is possibile that the southward‐
zonally elongated signature, covering about 50° in longitu- penetrating midlatitude Rossby wave and the QBWO orig-
dinal extent. As for wave perturbations near the equator, inating from the tropics jointly intensify the 10–20 day
roughly symmetric circulation centers exist in both hemi- oscillation amplitude over the East Asian extratropical
spheres, although the southern components are weaker and region. The concurrent contribution to 10–20 day oscillation
shifted eastward relative to those in the Northern Hemi- from the tropics and midlatitude still requires more inves-
sphere. The vertical structure along the cross section of the tigation to identify the interactions and their effect on the
QBWO perturbation centers tilts northwestward with height, QBWO.
giving rise to the vertical baroclinic structure. The rising
(sinking) motions accompanied by warm (cold) temperature
anomalies are located at the middle level, which are tightly [44] Acknowledgments. We would like to thank Miss S.‐Y. Lee and
Yu‐Chia Peng for assisting the initial stage of the analysis. Guanghua Chen
coupled with the positive (negative) vorticity north of 20°N was supported by the National Key Technology R&D Program (grant
and lead the vorticity by one eighth to one quarter of a 2008BAK50B02), the National Natural Science Foundation of China (grant
wavelength in the tropics. These signatures imply that the 40905024), and a visiting grant from the National Science Council (NSC).
C.‐H. Sui was supported by NSC grant 95‐2111‐M‐008‐003.
QBWO origin may be related to one of the equatorial wave
types with a similar period.
[41] Unlike other studies, the energetic analysis reveals
the underlying mechanism by which the QBWO develops References
and maintains. The energy balance averaged over a com- Aiyyer, A. R., and J. Molinari (2003), Evolution of mixed Rossby‐gravity
plete cycle of QBWO suggests that the waves are main- waves in idealized MJO environments. J. Atmos. Sci., 60, 2837–2855.
Bretherton, C. S., M. Widmann, V. P. Dymnikov, J. M. Wallace, and I. Blade
tained through baroclinic conversion north of 25°N and (1999), The effective number of spatial degrees of freedom of a time‐
diabatic latent heating in the tropics. The barotropic con- varying field. J. Clim., 12, 1990–2009.
version process for the generation of the EKE accounts for a Chang, C. P. (1970), Westward propagating cloud patterns in the tropical
Pacific as seen from time‐composite satellite photographs. J. Atmos.
smaller portion of the total energy source. In summary, the Sci., 27, 133–138.
energy transfer can be simply described by the following Chang, C. P., J. M. Chen, P. A. Harr, and L. E. Carr (1996), Northwestward
process: diabatic heating in the tropics and baroclinic con- propagating wave patterns over the tropical western North Pacific during
version in the subtropics jointly convert energy to EAPE, summer. Mon. Weather Rev., 124, 2245–2266.
Chen, G. H., and R. H. Huang (2009), Interannual variation of the mixed
which is further converted to EKE, through the negative Rossby‐gravity waves and their impact on tropical cyclogenesis in the
correlation of perturbation temperature and pressure vertical western North Pacific. J. Clim., 22, 535–549.
velocity. The EKE produced by conversion from EAPE and Chen, T. C., and J. M. Chen (1993), The 10–20‐day mode of the 1979
indian monsoon: Its relation with the time variation of monsoon rainfall.
barotropic processes can be redistributed by the conver- Mon. Weather Rev., 121, 2465–2482.
gence of eddy geopotential flux and advection by mean and Chen, T. C., and J. M. Chen (1995), An observational study of the South
perturbation wind. China Sea monsoon during the 1979 summer: Onset and life cycle.
Mon. Weather Rev., 123, 2295–2318.
[42] A detailed comparison between the observed wave Duchon, C. E. (1979), Lanczos filtering in one and two dimensions.
propagation and the theoretical estimate is made to verify J. Appl. Metereol., 18, 1016–1022.
that the origin of the QBWO is related to one kind of Dunkerton, T. J., and M. P. Baldwin (1995), Observation of 3–6‐day
equatorial wave. On the basis of the lag correlation, this meridional wind oscillations over the tropical Pacific, 1973–1992: Hori-
zontal structure and propagation. J. Atmos. Sci., 52, 1585–1601.
study reports that the tropical QBWO origin is identified Frank, W. M., and P. E. Roundy (2006), The role of tropical waves in tropical
with a zonal wave number of 6, and the westward phase cyclogenesis. Mon. Weather Rev., 134, 2397–2417.
speed and group velocity are consistent with an n = 1 ER Fukutomi, Y., and T. Yasunari (1999), 10–25 day intraseasonal variations
of convection and circulation over East Asia and western North Pacific
wave. A case study of the event during August 2002 during early summer. J. Meteorol. Soc. Jpn., 77, 753–769.
described the evolution of an n = 1 ER wave with a sym- Fukutomi, Y., and T. Yasunari (2002), Tropical‐extratratropical interaction
metric vortex about the equator into an off‐equatorial 10– associated with the 10–25 day oscillation over the western Pacific during
20 day oscillation, further supporting the interpretation. northern summer. J. Meteorol. Soc. Jpn., 80, 311–331.

13 of 14
D14113 CHEN AND SUI: QBWO OVER THE WESTERN NORTH PACIFIC D14113

Kalnay, E., et al. (1996), The NCEP/NCAR 40‐year reanalysis project. Reed, R. J., and E. E. Recker (1971), Structure and properties of synoptic‐
Bull. Am. Meteorol. Soc., 77, 437–471. scale wave disturbances in the equatorial western Pacific. J. Atmos. Sci.,
Kang, I. S., C. H. Ho, and Y. K. Lim (1999), Principal modes of climato- 28, 1117–1133.
logical seasonal and intraseasonal variations of the Asian summer mon- Reed, R. J., E. Klinker, and A. Hollingsworth (1988), The structure and
soon. Mon. Weather Rev., 127, 322–340. characteristics of African easterly wave disturbances as determined from
Kikuchi, K., and B. Wang (2009): Global perspective of the quasi‐ the ECMWF operational analysis/forecast system. Meteor. Atmos. Phys.,
biweekly oscillation. J. Climate, 22, 1340–1358. 38, 22–33.
Kiladis, G. N., and M. Wheeler (1995), Horizontal and vertical structure of Sobel, A. H., and C. S. Bretherton (1999), Development of synoptic‐scale
observed tropospheric equatorial Rossby waves. J. Geophys. Res., 100, disturbances over the summertime tropical northwest Pacific. J. Atmos.
22981–22997. Sci., 56, 3106–3127.
Kistler, R., et al. (2001), The NCEP/NCAR 50‐year reanalysis: Monthly Straub, K. H., and G. N. Kiladis (2002), Observations of a convectively
means CD‐ROM and documentation. Bull. Am. Meteorol. Soc., 82, coupled Kelvin wave in the eastern Pacific ITCZ. J. Atmos. Sci., 59,
247–267. 30–53.
Ko, K.‐C., and H.‐H. Hsu (2006), Sub‐monthly circulation features asso- Straub, K. H., and G. N. Kiladis (2003), Interactions between the boreal
ciated with tropical cyclone tracks over the East Asian monsoon area summer intraseasonal oscillation and higher‐frequency tropical wave
during July–August season. J. Meteorol. Soc. Jpn., 84, 871–889. activity. Mon. Weather Rev., 131, 945–960.
Ko, K.‐C., and H.‐H. Hsu (2009), ISO modulation on the submonthly wave Takayabu, Y. N., and T. Nitta (1993), 3–5 day‐period disturbances coupled
pattern and recurving tropical cyclones in the tropical western North with convection over the tropical Pacific Ocean. J. Meteorol. Soc. Jpn.,
Pacific. J. Clim., 22, 582–599. 71, 221–246.
Kuo, H. C., J. H. Chen, R. T. Williams, and C. P. Chang (2001), Rossby Wang, B., and X. Xie (1996), Low‐frequency equatorial waves in vertically
waves in zonally opposing mean flow: Behavior in northwest Pacific sheared zonal flow. Part I: Stable waves. J. Atmos. Sci., 53, 449–467.
summer monsoon. J. Atmos. Sci., 58, 1035–1050. Webster, P. J., and H. R. Chang (1988), Energy accumulation and emana-
Lau, K. H., and N. G. Lau (1990), Observed structure and propagation tion regions at low latitudes: Impacts of a zonally varying basic state.
characteristics of tropical summertime synoptic‐scale disturbances. J. Atmos. Sci., 45, 803–829.
Mon. Weather Rev., 118, 1888–1913. Wen, M., and R. H. Zhang (2008), Quasi‐biweekly oscillation of the con-
Lau, K. M., and S. Yang (1996), Seasonal variation abrupt transition, and vection around Sumatra and low‐level tropical circulation in boreal
intraseasonal variability associated with the Asian summer monsoon in spring. J. Clim., 136, 189–205.
the GLAGCM. J. Clim., 9, 965–985. Wheeler, M., G. N. Kiladis, and P. J. Webster (2000), Large‐scale dynamical
Li, C. Y., and Y. P. Zhou (1995), On quasi‐two‐week (10–20‐day) oscil- fields associated with convectively coupled equatorial waves. J. Atmos.
lation in the tropical atmosphere. Chin. J. Atmos. Sci., 19, 435–444 (in Sci., 57, 613–640.
Chinese). Xie, X., and B. Wang (1996), Low‐frequency equatorial waves in vertically
Liebmann, B., and H. H. Hendon (1990), Synoptic‐scale disturbances near sheared zonal flow. Part II: Unstable waves. J. Atmos. Sci., 53, 3589–
the equator. J. Atmos. Sci., 47, 1463–1479. 3605.
Liebmann, B., and C. A. Smith (1996), Description of a complete (interpo- Yanai, M., T. Maruyama, T. Nitta, and Y. Hayashi (1968), Power spectra of
lated) OLR dataset. Bull. Am. Meteorol. Soc., 77, 1275–1277. large‐scale disturbances over the tropical Pacific. J. Meteorol. Soc. Jpn.,
Madden, R. A., and P. R. Julian (1994), Observations of the 40–50‐day 46, 308–323.
tropical oscillation—A review. Mon. Weather Rev., 122, 814–837. Zangvil, A. (1977), On the presentation and interpretation of spectra of
Maloney, E. D., and M. J. Dickinson (2003), The intraseasonal oscillation large‐scale disturbances. Mon. Weather Rev., 105, 1469–1472.
and the energetics of summertime tropical western North pacific synoptic‐ Zhang, C., and P. J. Webster (1989), Effects of zonal flows on equatorially
scale disturbances. J. Atmos. Sci., 60, 2153–2168. trapped waves. J. Atmos. Sci., 46, 3632–3652.
Maloney, E. D., and D. L. Hartmann (2001), The Madden–Julian Oscillation,
barotropic dynamics, and North Pacific tropical cyclone formation. Part I: G. Chen, Center for Monsoon System Research, Institute of Atmospheric
Observations. J. Atmos. Sci., 58, 2545–2558. Physics, Chinese Academy of Sciences, PO Box 2718, Beijing 100190,
Matsuno, T. (1966), Quasi‐geostrophic motions in the equatorial area. China. (cgh@mail.iap.ac.cn)
J. Meteorol. Soc. Jpn., 44, 25–43.
C.‐H. Sui, Institute of Hydrological and Oceanic Sciences and
Molinari, J., K. Lombardo, and D. Vollaro (2007), Tropical cyclogenesis Department of Atmospheric Sciences, National Central University, 300
within an equatorial Rossby wave packet. J. Atmos. Sci., 64, 1301–1317. Jhongda Rd., Jhongli City, Taoyuan County 320, Taiwan.

14 of 14

You might also like