1. Introduction
The multiscale nature of circulation in the tropical western North Pacific (WNP) has become an interesting subject in recent years (e.g., Holland 1995; Hsu 2005). These multiscale features include monsoon circulation, intraseasonal oscillation (ISO), 3–8-day synoptic waves, and tropical cyclones (TCs). It has been reported in numerous studies that TCs tend to form and develop in the favorable condition provided by the large-scale tropical circulation (e.g., Carr and Elsberry 1995; Chang et al. 1996; Davidson and Hendon 1989; Dickinson and Molinari 2002; Frank and Roundy 2006; Fu et al. 2007; Ge et al. 2007; Krouse et al. 2008; Li and Fu 2006; Li 2006; Shapiro and Ooyama 1990; Sobel and Bretherton 1999). Our previous study (Ko and Hsu 2006, hereafter KH06) discovered a 7–30-day wave pattern that propagates north-northwestward from the northeast of Papua New Guinea to the area between Taiwan and Japan. The pattern was most active during mid to late summer (July–August) and closely tied to the fluctuation of the monsoon trough/subtropical high system in the WNP. More than 70% of the identified cases were associated with at least one recurving tropical cyclone when the cyclonic circulation in the pattern reached its peak phase near the East China Sea. The cyclonic circulation created a favorable environment (e.g., cyclonic background circulation, stronger moisture convergence, etc.) for recurving tropical cyclones to form, develop, and move along with this wave pattern. This pattern was named the TC/submonthly wave pattern because of its close coupling with TCs and a time scale (7–30 days) longer than the synoptic perturbation.
The ISO in the WNP has a modulating effect on the synoptic system and the TC/submonthly wave pattern mentioned above (e.g., Maloney and Dickinson 2003; Liebmann et al. 1994; Ko and Hsu 2009, hereafter KH09). KH09 examined the ISO modulating effect on the TC/submonthly wave pattern. Their results indicated that, in the ISO westerly phase, the TC/submonthly wave pattern was better organized in the cyclonic and moisture-convergence environment of a strong eastward extending monsoon trough, whereas the pattern was loosely organized in the easterly phase when the monsoon trough was weak and retreated westward. It was also shown in their study that the TCs associated with the submonthly wave pattern were more energetic in the ISO westerly phase than in the easterly phase. While the circulation characteristics of the TC/submonthly wave pattern were well explored by KH09, the propagation and maintenance mechanisms were still poorly understood.
Explaining the ISO modulating effect on the tropical disturbances in the WNP in terms of the barotropic dynamics has attracted more attention in the last decade (Hartmann and Maloney 2001; Maloney and Hartmann 2001; Aiyyer and Molinari 2008; Chen and Sui 2010; Hsu et al. 2011). Maloney and Hartmann computed and showed that the westerly jet in the westerly MJO periods provided more kinetic energy for the tropical disturbances to grow. Hsu et al. (2011) also found that the enhanced barotropic energy conversion favored the generation and development of synoptic seed disturbances for cyclogenesis during the ISO active phase. The upscale influence of synoptic perturbations and TCs on the large-scale circulations has attracted more attention in recent years. Hsu et al. (2008a,b) adopted the procedures developed by Kurihara et al. (1993, 1995) and Wu et al. (2002) in removing TCs from the analysis data. They conducted a series of statistical studies and found that TCs contributed significantly to the long-term mean, seasonal mean, and intraseasonal and interannual variance in the 850-hPa vorticity along the TC tracks in the tropical WNP. It was demonstrated that the TC effect could not be removed by long-term averaging and low-pass filtering. A similar upscale feedback from synoptic perturbation to the ISO was also reported by Zhou and Li (2010) and Hsu et al. (2011).
Although the TC effect on the submonthly wave pattern was briefly investigated by KH09, a quantitative assessment was not carried out. In view of the coexistence of TCs and the submonthly wave pattern, an exploration of the TC effect on the submonthly wave pattern should yield valuable information on multiscale interaction in the WNP. The objectives of this study are to 1) explore the propagation and maintenance mechanisms of the submonthly wave pattern and 2) estimate the TC contribution to the mechanisms. Following the previous studies by KH06, KH09, and Hsu et al. (2008a,b), this study pays more attention to vorticity and kinetic energy analysis at 850 hPa. One of the major reasons is that the TC removal procedure was only applied to the 850-hPa wind where the TC vortex in the global analysis was close to the observed TC center. Furthermore, the TC/submonthly wave pattern exhibited largest amplitude in the lower troposphere. Section 2 describes the data and analysis procedures. The results are divided into three parts. The structure and vorticity budget of the TC/submonthly wave pattern are shown in section 3. The kinetic energy and various conversion terms are discussed in section 4. The TC feedback is explored in section 5. Conclusions and discussion are presented in section 6.
2. Data and analysis procedures
The data used in this study were extracted from the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) (Uppala et al. 2005). The ERA-40 contains 6-hourly (0000, 0600, 1200, and 1800 UTC) temperature, humidity, horizontal winds, vertical p velocity, and geopotential on a 2.5° × 2.5° latitude–longitude grid. This study analyzed the 23-yr data during June–October (JJASO) 1979–2001. A Butterworth band-passed filter (Kaylor 1977; Hamming 1989), following Hsu and Weng (2001), KH06, and KH09, was applied to extract the 7–30-day (submonthly) and 30–80-day (intraseasonal) fluctuations.
A TC removal routine proposed by Kurihara et al. (1993, 1995) in the Geophysical Fluid Dynamics Laboratory hurricane prediction system and Wu et al. (2002) in typhoon simulation was applied to the 850-hPa wind field of the ECMWF analysis to remove the TC wind from the global analysis. This TC-removed dataset is the same as that used in Hsu et al. (2008b). Readers are referred to their papers for details. The procedure is described briefly as follows. The wind fields were decomposed into the basic and disturbance fields. Tangential winds of the TCs (based on the Joint Typhoon Warning Center at Guam) were then detected and removed from the disturbance field and replaced by interpolated winds. The TC-removed disturbance field and the interpolated winds were then summed to create the non-TC component, which was then added back into the basic field to form the environmental flow.
The cases selected in this study are the same as those in KH09. The criteria are described briefly here: The submonthly cases were selected based on the 7–30-day filtered time series of 850-hPa wind speed averaged over the base region between Taiwan and Japan (20°–35°N, 125°–140°E) where a variance maximum of wind speed existed. The submonthly cases were selected when the positive anomalous maxima were greater than 1 m s−1, which was equivalent to 0.75 standard deviation. A TC/submonthly case was further selected when at least a TC appeared in the base region (with a 1.5° latitude–longitude buffer zone just outside the edge) between day −1 and day +1 of a selected submonthly case. The TC/submonthly cases were further classified into two categories: ISO westerly and easterly phases. A TC/submonthly case was classified as a case in the westerly (easterly) phase if the anomalous 30–80-day filtered zonal wind averaged over the ISO base region (5°–15°N, 100°–125°E), where maximum variance in the intraseasonal zonal wind was located, at the occurrence time of the submonthly case was greater than 1.5 m s−1 (less than −1.5 m s−1). The ±1.5 m s−1 thresholds were approximately ±0.5 standard deviation of the 30–80-day filtered zonal wind time series. The resultant number of submonthly cases is 40 for the ISO westerly phase and 24 for the ISO easterly phase.
3. Vorticity budget
To demonstrate the horizontal structure of the background circulation features in the lower troposphere, the 25-day mean 850-hPa wind and vorticity for both ISO westerly and easterly phases are shown in Figs. 1a and 1b. The mean flow pattern in the westerly phase exhibited a strong monsoon trough, characterized by cyclonic circulation and a positive vorticity maximum between 10° and 25°N extending eastward from the northern South China Sea (SCS) to the Philippine Sea. Farther to the southeast of the monsoon trough was a confluent zone, a positive vorticity zone of smaller amplitude. To the south and north of the positive vorticity zone were two negative vorticity zones. As will be revealed later, the strong southeasterly between the positive and negative vorticity zones in the Philippine Sea was an important background circulation feature that significantly steered the TC/submonthly wave pattern propagation. In contrast to the westerly phase, the easterly phase is characterized by a weak and westward-retreating monsoon trough. As a result, the relative vorticity in the northern SCS and the western Philippine Sea was almost zero, while a weak positive vorticity belt still existed south of the subtropical Pacific anticyclone. The prevailing southeasterly in the Philippine Sea seen in the westerly phase no longer existed; instead, a weak southwesterly flow prevailed along the East Asian coast. As will be shown below, the different background circulation configuration between the westerly and easterly phase results in the distinct structure and propagation characteristics of the TC/submonthly wave pattern. The mean flow for each ISO phase is calculated individually and used in the energetic and vorticity budget term calculation for both TC and TC-removed cases.
(a) Westerly and (b) easterly climatological vorticity at 850 hPa averaged over 25 days for westerly and easterly cases. The contour interval is 3(10−6 s−1). Also shown are the wind vectors. Wind speeds ≤2 m s−1 are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
a. Propagation tendency
Figure 2 shows the composite maps of the 7–30-day filtered 850-hPa vorticity and wind field. At day −3 in the westerly phase (Fig. 2a) a well-organized wave pattern existed in the WNP. The pattern was oriented in the northwest–southeast direction from the equatorial WNP toward the area between Taiwan and Japan, while the individual vorticity perturbation tilted roughly in the northeast–southwest direction. Stronger wind vectors (exceeding 95% confidence level) were evident surrounding the cyclonic circulation in the wave pattern. The wave pattern propagated northwestward toward eastern China in the next few days when both the vorticity and wind reached their maximum phase (Fig. 2c). After day 0 the wave pattern continued propagating northwestward and started decaying (Fig. 2e).
(a),(c),(e) Westerly and (b),(d),(f) easterly composites of 7–30-day filtered vorticity at 850 hPa. The contour interval is 3(10−6 s−1) and areas exceeding the 95% confidence level are shaded: positive (yellow) and negative (blue). Also shown are the wind vectors exceeding the 95% confidence level. The zero contours are omitted. The thick dashed lines are chosen for the cross sections that will be shown later.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
In contrast to the evidently wavy structure in the westerly phase, the day −3 pattern in the easterly phase (Fig. 2b) was confined to a much smaller area near Taiwan and Japan and exhibited less wavelike structure (KH09). A slightly better organized wave pattern emerged at day 0 (Fig. 2d). In the next few days, the perturbation in the eastern part of the wave pattern moved northward toward Japan, while the one on the western end moved northwestward toward southern China (Fig. 2f).
b. Westerly phase
The advection terms are examined below to reveal the relative influence of various advection processes on the propagation tendency of the TC/submonthly wave pattern. Shown in Fig. 3 are the advection terms at day 0 in the westerly phase. The beta effect term (Fig. 3a), which resulted in positive and negative vorticity tendency in the western and eastern part of the positive vorticity anomaly south of Japan, had a northwestward steering effect on the vorticity anomalies over the Philippine Sea because of the northeast–southwest tilting of the individual circulation circuit.
The 850-hPa eddy vorticity terms (×10−11 s−2, shaded) at day 0 in the westerly phase for the (a) beta effect, (b) eddy advection, (c) mean flow advection, (d) nonlinear term, and (e) stretching term. Also shown is the 7–30-day composite vorticity (×10−6 s−1, contour interval: 3) and the zero contours are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
The eddy advection term (Fig. 3b) exhibited a more complex pattern due to the relative position of the wind perturbation and background vorticity distribution (Fig. 1a). A major eddy vorticity maximum, that is, the primary one near the core of the cyclonic anomaly, was found south of Japan. This primary vorticity anomaly was located to the northwest of the positive background vorticity center in the Philippine Sea. The advection effect of the anomalous cyclonic circulation (Fig. 2c) associated with positive anomalies therefore resulted in a negative (positive) vorticity tendency northwest (southeast) of the primary vorticity anomaly. This advection effect explains the quadrupole structure of the vorticity tendency surrounding the positive vorticity anomaly at day 0. Such a tendency distribution would tend to move the primary vorticity anomaly southeastward. Since the eddy advection term outweighed the beta effect, the total effect of the eddy advection term would prevent the primary vorticity anomaly from propagating northwestward.
The mean flow advection term (Fig. 3c) seemed to play a dominant role in steering the wave pattern toward the northwest. The primary positive vorticity anomaly was located in the background southeasterly to the northeast of the monsoon trough (Fig. 1a). The advection by the southeasterly, which resulted in the positive (negative) tendency in the northwest (southeast) of the major vorticity anomaly, tended to move the positive and negative anomalies northwestward. The nonlinear eddy advection (Fig. 3d) resulted in a larger negative tendency in the southern half of the primary vorticity anomaly and scattering of a smaller positive tendency in the northern half. Since the negative tendency in the south had larger amplitude and was better organized than its counterpart in the north, the nonlinear eddy term did not contribute to the northwestward propagation of the primary positive vorticity anomaly. Instead, it seemed to damp the positive vorticity anomaly while moving the negative vorticity anomaly northwestward. The positive tendency contributed by the stretching term (Fig. 3e) was collocated with the primary positive vorticity anomaly but with larger amplitude in the south. The collocation between the stretching term and the primary positive vorticity anomaly suggests that the stretching term acted mainly to enhance the amplitude of the vorticity anomaly but contributed little to the propagation.
The northwest–southeast (Fig. 2c) vertical cross sections of perturbation vorticity along with the budget terms at day 0 in the westerly phase are shown in Fig. 4. The vertical structure of the vorticity anomaly exhibited a barotropic structure below 200 hPa and maximized in the lower troposphere. Superimposed on the vorticity anomaly were the various terms contributing to the vorticity tendency. The beta effect and mean flow advection contributed to the northwestward propagation (Figs. 4a,c). The nonlinear advection tended to damp the positive vorticity in addition to relatively weak contribution to the propagation (Fig. 4d). Conversely, the eddy relative vorticity advection acted to move the vorticity in the opposite direction (Fig. 4b). Among all advection terms, the mean flow advection term contributed dominantly to the northwestward propagation of the pattern throughout the troposphere, while the beta term was largely canceled by the eddy relative vorticity advection term.
Cross sections (see Fig. 2) of eddy vorticity terms (×10−11 s−2, shaded) at day 0 in the westerly phase for the (a) beta effect, (b) eddy advection, (c) mean flow advection, and (d) nonlinear term. Also shown is the 7–30-day composite vorticity (×10−6 s−1, contour interval: 3) and the zero contours are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
c. Easterly phase
As seen in Figs. 2b, 2d, and 2f, the primary positive vorticity anomaly in the easterly phase propagated essentially northward from day −3 to day 3. The 850-hPa vorticity budget terms, shown in Fig. 5, demonstrate what is responsible for this northward propagation. Similar to that as in the westerly phase, the beta effect (Fig. 5a) basically helped the vorticity anomaly to move westward. The eddy relative vorticity advection term (Fig. 5b) is small because the wind anomaly (Fig. 2d) occurred mainly in the region of small background relative vorticity (Fig. 1b). As a result, the beta effect (Fig. 5a) dominated the total eddy vorticity advection term. While the eddy advection term tended to move the positive vorticity anomaly westward, the mean flow advection (Fig. 5c) contributed to the northward propagation tendency because of the dominant southerly flow off the East Asian coast. The nonlinear eddy advection term (Fig. 5d) exhibited a positive tendency in the east and northwest of the positive vorticity anomaly and a negative tendency in the southwest and northeast. This pattern acted to squeeze the positive vorticity anomaly. The stretching term (Fig. 5e), though stronger than the other terms, tended to enhance the vorticity anomaly but contribute little to the propagation.
The 850-hPa eddy vorticity terms (×10−11 s−2, shaded) at day 0 in the easterly phase for the (a) beta effect, (b) eddy advection, (c) mean flow advection, (d) nonlinear term, and (e) stretching term. Also shown is the 7–30-day composite vorticity (×10−6 s−1, contour interval: 3) and the zero contours are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
The vertical structure along 130°E (Fig. 2d) of the perturbation vorticity at day 0 in the easterly phase, shown in Fig. 6, illustrates the major processes leading to the propagation. Compared to that in the westerly phase, the vertical cross section of vorticity presented in Fig. 6 exhibits a weaker and more concentrated maximum centered near 25°–30°N. The beta effect (Fig. 6a) exhibits a broad area of weak negative tendency because it basically contributed to westward propagation. The eddy relative vorticity advection resulted in a negative vorticity tendency to the north of the positive vorticity anomaly (Fig. 6b), which was against the northward propagation tendency of the pattern. This eddy advection pattern could also be inferred from the perturbation wind (Fig. 2) and mean vorticity fields (Fig. 1). The mean flow advection term (Fig. 6c) dominated in the total advection term and the nonlinear eddy term (Fig. 6d) showed a weaker northward tendency near the vorticity maximum. The aforementioned advection terms indicate that the mean flow advection was also responsible for the northward propagation of the vorticity pattern as in the westerly phase.
Cross sections (along 130°E) of eddy vorticity terms (×10−11 s−2, shaded) at day 0 in the easterly phase for the (a) beta effect, (b) eddy advection, (c) mean flow advection, and (d) nonlinear term. Also shown is the 7–30-day composite vorticity (×10−6 s−1, contour interval: 3) and the zero contours are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
4. Kinetic energy and barotropic conversion
Figures 7a and 7b present the vertically averaged PKE (25-day averages) for both westerly and easterly phases. The maximum areas of PKE were mostly located in midlatitudes centered around 45°N, 165°E. A branch of local maximum PKE near 130°E in the westerly phase (Fig. 7a), collocated with the axis of the TC/submonthly wave pattern, extended southward from 30°N, 130°E to 15°N, 130°E. Such a southward-extending PKE maximum was less evident in the easterly phase (Fig. 7b). The north–south cross sections of the composite PKE averaged over the axis of the maximum PKE (125°–135°E) are shown in Figs. 7c and 7d. The PKE revealed maxima near 15°–30°N at the lower level of the troposphere in the westerly phase and this PKE pattern was stronger than that in the easterly phase. This subtropical maximum is well separated from the large PKE in the extratropical upper troposphere.
Composite vertically averaged perturbation kinetic energy for the (a) westerly and (b) easterly phases and (c),(d) the corresponding cross sections (averaged over 125°–135°E). The contour interval is 4 m2 s−2 and areas greater than 16 m2 s−2 are shaded.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
The vertical cross sections (averaged over 125°–135°E) of major energetic budget terms in Eq. (2) and the diabatic term in Eq. (4) during the westerly phase are shown in Fig. 8. A maximum of positive barotropic energy conversion (Fig. 8a) was located below 400 hPa between 10° and 25°N with the maximum in the lower troposphere. Overlaying this positive anomaly is a broad area of negative values. The baroclinic energy conversion term (Fig. 8b) and the diabatic term (Fig. 8d) are collocated and both maximized around 250 hPa. This collocation indicated that the PAPE generated by the diabatic heating was converted to PKE mostly in the upper troposphere. This collocated distribution was also identified in the synoptic tropical disturbances (e.g., Maloney and Dickinson 2003). This process occurred mainly in the upper troposphere and, therefore, could not explain the kinetic energy distribution, which was located below 400 hPa and peaked in the lower troposphere. Instead, the barotropic energy conversion from the background flow to perturbations (Fig. 8a) was the major kinetic energy source for the development of the wave pattern. Although the kinetic energy generation term (Fig. 8c) exhibited large maxima in the lower and upper troposphere, it only partially cancelled the barotropic energy conversion term through a local negative maximum between 15° and 25°N. The energetic budget terms in the easterly phase exhibited a similar distribution as in the westerly phase but with much smaller amplitudes (not shown).
Westerly composite cross sections (averaged over 125°–135°E) of the (a) barotropic, (b) baroclinic and (c) generation terms as in Eq. (2), and (d) the diabatic heating term as in Eq. (4). The contour interval is 20(×10−5) m2 s−3. The zero contours are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
These results suggested the dominance of the barotropic energy conversion term in the energy conversion from the background intraseasonal flow to perturbations. Since the amplitude of the wave pattern and the dominant energy conversion is maximized in the lower troposphere, the following analysis will focus on 850 hPa to explore the barotropic perspective of kinetic energy conversion between the intraseasonal mean flow and the TC/submonthly wave pattern.
a. Barotropic energy conversion in the westerly phase
The PKE map (Fig. 9a) for the westerly phase revealed the existence of a PKE maximum area exceeding the 95% confidence level between Taiwan and Japan where the TC/submonthly wave pattern was most active. The barotropic energy conversion term (Fig. 9b), which represented the kinetic energy exchange between the westerly-phase ISO circulation and the TC/submonthly wave pattern, also exhibited a positive barotropic conversion maximum to the southeast of the PKE maximum in the Philippine Sea.
Composites of (a) PKE, (b) barotropic term, (c) 
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
Figure 9c shows that the northeastern corner of the westerly tongue, where both 
A comparison between the anomalous wind vectors and background westerly distribution yields interesting insight about the eddy–mean flow interaction in this region. The discussion in the previous paragraph explained why the TC/submonthly wave pattern tended to occur between the monsoon trough and the anticyclonic ridge and why each circulation circuit was oriented in the northeast–southwest direction (Fig. 2c). Eddies elongated in such an orientation could efficiently extract the kinetic energy from the mean flow to sustain growth in the region of large mean flow gradient (Figs. 9c,d). This perfect match is, indeed, what we have been observing in the western North Pacific during boreal summer.
Such a background configuration was also favorable for the propagation of the TC/submonthly wave pattern. The northeast–southwest orientation of the eddy, which was almost orthogonal to the mean southeasterly (Fig. 1a), was likely the optimal orientation to maximize the mean flow advection effect. However, the propagation of eddies would be slower than the speed of the southeasterly. This was because the anomalous eddy flow would create significant eddy advection of mean vorticity (Fig. 3b) to counteract the mean flow advection.
This discussion suggests, in a barotropic point of view, that the TC/submonthly wave pattern with a circulation circuit in the northeast–southwest orientation was the one to have optimal growth and propagation in the confluent zone between the monsoon trough and the anticyclonic ridge where kinetic energy could be extracted most efficiently. This is likely the reason why the TC/wave pattern with the specific configuration reported here is frequently observed in the WNP during boreal summer.
b. Barotropic energy conversion in the easterly phase
The kinetic energy and barotropic conversion distribution of the TC/submonthly wave pattern during the ISO easterly phase are shown in Figs. 10a and 10b. The amplitudes of PKE and barotropic conversion were about one-half and one-quarter of their westerly-phase counterparts, respectively. Additionally, most of the areas in Figs. 10a and 10b failed to make the 95% confidence level. The maximum PKE was confined between Taiwan and Japan and the maximum barotropic conversion of a much smaller area was found in the northeast of the Philippines. The locations of the relatively weaker PKE and barotropic conversion were consistent with the observation that the genesis and propagation of the TC/submonthly wave pattern in the easterly phase all occurred near the East Asian coast. Figures 10c and 10d present the background 850-hPa zonal wind, 
Composites of (a) PKE, (b) barotropic term, (c) 
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
The weak zonal wind gradient associated with the weaker and westward-withdrawn westerly tongue in the ISO easterly phase resulted in the confinement of the energy-extracting region near the East Asian coast. The zonally elongated eddies (like those shown in Fig. 2) are the ones that were favorable to grow and propagate northward in this region in view of the zonal wind distribution and the prevailing background southerly (Fig. 1b). If northeast–southwest oriented eddies as in the westerly phase appeared in the Philippine Sea, they would face an environment less favorable for further growth and propagation because of the weak 
A sensitivity test was done by switching the background flow patterns between the westerly and easterly phase. That is, the barotropic conversion terms in the westerly (easterly) phase were calculated based on the westerly (easterly) phase perturbations and the easterly (westerly) phase background flow. The results (not shown) indicated that both background flow and anomalies contributed to the amplitudes of energy conversion. Hypothetically, if the easterly perturbations were put in the westerly background flow, it would convert more kinetic energy than the observed easterly perturbation/easterly mean pair but less than the observed westerly perturbation/westerly mean pair. Conversely, the hypothetically westerly perturbations in the easterly background flow would convert more kinetic energy than the observed easterly perturbation/easterly mean pair but less than the observed westerly perturbation/westerly mean pair. Results of this comparison demonstrated the positive relationship between the background flow and perturbations: that is, stronger background flow produced stronger wave activity through a more vigorous eddy-mean flow interaction. It should be cautioned that the aforementioned results do not suggest that energy conversion is more efficient in one particular eddy–background flow combination than the others.
5. TC feedback
a. Propagation
Whether and how the tropical cyclones affected the propagation of the wave pattern is explored in this section. The TC-removed vorticity fields at day 0 are shown in Fig. 11. The vorticity near the center of the cyclonic circulation at day 0 in the westerly phase decreased by about 50% after removing TCs (Fig. 11a). Although the TC removal resulted in decreased amplitude, it hardly changed the overall structure, scale, and propagation tendency of the wave pattern (Figs. 3 and 11).
The 850-hPa eddy vorticity terms (×10−11 s−2, shaded) at day 0 for the TC-removed data in the westerly phase for the (a) beta effect, (b) eddy advection, (c) mean flow advection, and (d) nonlinear term. Also shown is the 7–30-day composite vorticity (×10−6 s−1, contour interval: 3) and the zero contours are omitted.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
Also shown in Fig. 11 are the various vorticity budget terms for the TC-removed 850-hPa field in the westerly phase. The major difference between the original (Fig. 3) and TC-removed fields (Fig. 11) is again the reduction of the amplitudes while the spatial distribution patterns remain similar for every term except the nonlinear eddy term. It is not surprising because in the calculation TCs affected only the eddy amplitudes and the long-term mean flow was kept the same as in the original field during the calculation. A test was done to check the results using the TC-removed mean flow instead of the original mean flow. During the test, the mean flow was calculated from the TC-removed 850-hPa wind field. As pointed out in Hsu et al. (2008b), TCs contributed notably to the long-term mean circulation and variance but had a weaker effect on the spatial pattern. The test results indicated that using the TC-removed mean flow in the calculation only resulted in amplitude reduction in all linear terms. It follows that using the original mean flow does not affect our results notably.
Conversely, the TCs had a much larger effect on the nonlinear eddy term. The large positive and negative vorticity advections seen in Fig. 3d were no longer evident using the same plotting scheme. As noted above, the nonlinear eddy advection contributed much more to the negative vorticity tendency in the south than the positive tendency in the north of the primary positive vorticity anomaly and, therefore, had a squeezing effect on the primary positive vorticity anomaly to reduce the vortex meridional scale. Since the nonlinear term became negligible in the TC-removed fields, it suggested that TCs were the main cause of this squeezing effect. Most TCs reached peak strength before recurving toward higher latitudes (e.g., Evans and McKinley 1998; Harr 2010). It was therefore speculated that the amplitude reduction of TCs likely resulted in the squeezing effect on the large-scale vorticity anomaly discussed above because they would have a larger impact on the wave pattern in the south than in the north. The influence of removing TCs in the easterly phase showed a similar amplitude-decreasing pattern as in the westerly phase but with a larger reduction in amplitude near the center of the vorticity maximum (not shown).
b. Kinetic energy
The PKE and barotropic conversion of the TC-removed 850-hPa wind field in the ISO westerly phase are shown in Figs. 12a and 12b. While the patterns remained similar to those derived from the original fields (Figs. 9a and 9b), the amplitudes dropped by more than one-half for both PKE and barotropic conversion. This contrast revealed that TCs contributed more than one-half of the kinetic energy of the TC/wave pattern and enhanced the energy conversion from the mean flow to eddies by about the same amount contributed by the large-scale eddies. However, the reduction did not prevent PKE and barotropic conversion from exceeding the 95% confidence level. The reduction in the barotropic conversion in the TC-removed field was contributed by the first two terms of the barotropic conversion, 
Composites of (a) PKE, (b) barotropic term, (c) 
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
To quantify the extent of the TCs contribution to the eddy kinetic energy and conversion, the ratios of (Total − TCremoved)/Total PKE and the barotropic conversion term are presented in Fig. 13. More than 50% of the eddy kinetic energy in a large region was contributed by TCs for both ISO phases (Figs. 13a,c). The areas of decreasing kinetic energy were larger in the westerly than easterly phase over the WNP, indicating a greater spatial extent of the TC impact in the westerly phase. The (Total − TCremoved)/Total ratios of the barotropic conversion term for both ISO phases are shown in Figs. 13b and 13d. Since the mean-state flow was the same for the calculation with or without TCs, the ratios depended on the changes in the magnitude of 
Westerly composites for the ratios of the (Total − TCremoved)/Total fields of the (a) PKE and (b) total barotropic term and (c),(d) the corresponding easterly composites. The contour interval is 0.1 for PKE ratios and 0.2 for total barotropic term ratios.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
Since TCs were more active and tended to occur in clusters during the westerly phase than in the easterly phase, the significant contribution to both PKE and barotropic conversion occurred in a much larger spatial domain. The results presented in this section indicate that, while TCs were provided with a favorable development condition by the submonthly wave pattern, they in turn acted to enhance the overall kinetic energy of the TC/submonthly wave pattern and helped extract significantly more energy from the background ISO circulation. This effect was seen in a much larger area in the ISO westerly phase due to an enhanced monsoon trough. This result does not imply an increase in the energy conversion efficiency due to the presence of TCs. To accurately answer the question, the same calculation has to be done for the submonthly wave pattern with no embedded TCs. This interesting issue is beyond the scope of the present study.
6. Summary and concluding remarks
This study explored the propagation and maintenance mechanisms of the TC/submonthly wave pattern and assessed the TC contribution to these mechanisms during the intraseasonal oscillation westerly and easterly phases in the tropical werstern North Pacific. The TC/submonthly wave pattern exhibited equivalent barotropic vertical structure with a maximum in the lower troposphere. The vorticity budget was calculated to investigate the propagation tendency of the wave pattern at all levels from 1000 to 100 hPa during both ISO westerly and easterly phases. Results are summarized in the schematic diagrams shown in Figs. 14a and 14b. In the ISO westerly phase, the dominant effect of the background flow advection and beta effect tended to move the TC/submonthly wave pattern northwestward and westward, respectively, while the eddy relative vorticity advection acted against it. The nonlinear eddy advection had a weaker effect on propagation but tended to squeeze the perturbation vorticity meridionally by reducing the amplitude of the main vorticity anomaly in the south. In the easterly ISO phase, the mean flow advection tended to move the vorticity anomaly northward, whereas the beta effect tended to move the vorticity anomaly westward. The dominant effect of the mean flow advection over other advection terms resulted in the northward propagation. The stretching term in both phases acted mainly to enhance the amplitude of the vorticity anomaly in both phases and had a weaker effect on the propagation. The results indicate that the mean flow advection played the dominant role in TC/wave pattern propagation from 1000 to 400 hPa in both westerly and easterly phases. Stronger mean flow in the westerly phase resulted in a stronger propagation tendency of the TC/submonthly wave pattern than in the easterly phase.
Schematic diagrams for (a) westerly and (b) easterly phases at 850 hPa. The thin lines (solid for cyclonic and dashed for anticyclonic) with arrows represent the vorticity fields of the submonthly wave patterns. The shaded areas represent strong zonal wind (westerly) bands and the open arrows are the climatological flow near the Taiwan–Japan region. The individual effects (mean: 
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00643.1
A kinetic energy study was conducted to examine the maintenance mechanism. The wave pattern exhibited maximum kinetic energy in the lower troposphere. Among the four dominant terms in the energy budget equation, the barotropic energy conversion term was the only one that maximized in the mid to lower troposphere where the anomalous circulation of the submonthly wave pattern was located. Baroclinic conversion and diabatic heating, which were found in the upper troposphere, tended to cancel out and therefore contributed little to the maintenance of the wave pattern. These results indicate that the kinetic energy of the wave pattern was mainly maintained through the barotropic energy conversion from the background ISO flow.
In the westerly phase, a positive barotropic conversion maximum area existed to the southeast of the eddy kinetic energy maximum in the Philippine Sea (Fig. 14a). This region was located to the northeast of the background westerly maximum area in the southern flank of the monsoon trough, where the zonal and meridional gradients of the mean flow westerly were largest. The TC/submonthly wave pattern passed through this region and extracted energy from the mean flow with its southwest–northeast oriented eddy flows. This was likely the reason why the northwestward propagation of the northeast–southwest TC/wave pattern was frequently observed in the WNP during boreal summer because the wavelike perturbations in this configuration were more favorable to develop during the enhanced monsoon trough (e.g., the ISO westerly phase in this study). Eddies in other orientation would be much less favorable for further growth and even to decay (e.g., eddies oriented in northwest–southeast direction). Similar eddy–mean flow interaction occurred in the easterly phase. It was, however, much weaker because the barotropic energy conversion was confined near the East Asian coast because of the weaker and westward-withdrawn background westerly flow (Fig. 14b). The westward-retreated monsoon trough resulted in eddies of different orientation that propagated north-northeastward due to the prevailing south-southwesterly background flow.
A TC-removal technique was applied to the circulation data to study the feedback of TCs on the large-scale submonthly wave pattern during two extreme ISO phases. Removing TCs reduced the vorticity anomaly and budget terms by more than 50% in both phases. The nonlinear term, which almost vanished after removing the TCs, was affected most among all budget terms. This large effect is expected, considering the significant effect of TCs on eddy amplitude. While TCs had a strong feedback on the amplitude of the vorticity anomaly and budget, they had a smaller effect on the overall structure and propagation of the wave pattern.
Removing TCs also decreased the eddy kinetic energy and barotropic energy conversion by more than 50% in both phases but had a larger influence in the westerly phase in terms of the spatial extent. All of these results indicated that, while TCs developed in a favorable background flow provided by the submonthly wave pattern, they in turn enhanced the amplitudes of vorticity and kinetic energy of the TC/wave pattern and helped extract significantly more energy from the background ISO circulation. This feedback was stronger in the ISO westerly phase because of the stronger clustering effect of an enhanced monsoon trough on TCs.
This study focused only on the feedback of TCs to the submonthly wave pattern. Further studies on the feedback to the ISO and the moisture effect are underway to uncover the mechanism of the complex multiscale interaction over East Asian summer monsoon systems. This study revealed that TCs and the large-scale circulation in the WNP were mutually intertwined through rigorous eddy–mean flow interaction and therefore should be treated as an integrated multiscale system. It means that we need to understand the two-way interaction between different scales and take that into account to reach a more accurate interpretation of climate variability in the WNP. Another implication is that models need to properly simulate the interaction between various scales to correctly simulate the climate variability in the WNP.
Acknowledgments
The authors thank Miss Yan-Lan Wang and Dr. Chun-Hsiung Weng for computing the TC-removal circulation data. Comments from anonymous reviewers help improve the manuscript and the authors appreciate their suggestion and comments. This study was supported by the National Science Council, Taiwan ,under Grant NSC 99-2111-M-017-001-MY3 issued to Ken-Chung Ko and Grant NSC 98-2111-M-002-013-MY3 issued to Huang-Hsiung Hsu. Part of the work was done when K.-C. Ko was on sabbatical leave to Research Center for Environmental Changes of Academia Sinica and he is grateful to the support of the institution.
REFERENCES
Aiyyer, A., and J. Molinari, 2008: MJO and tropical cyclogenesis in the Gulf of Mexico and eastern Pacific: Case study and idealized numerical modeling. J. Atmos. Sci., 65, 2691–2704.
Briegel, L. M., and W. M. Frank, 1997: Large-scale influences on tropical cyclogenesis in the western North Pacific. Mon. Wea. Rev., 125, 1397–1413.
Carr, L. E., and R. L. Elsberry, 1995: Monsoonal interactions leading to sudden tropical cyclone track changes. Mon. Wea. Rev., 123, 265–289.
Chang, C.-P., J. M. Chen, P. A. Harr, and L. E. Carr, 1996: Northwestward-propagating wave patterns over the tropical western North Pacific during summer. Mon. Wea. Rev., 124, 2245–2266.
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.
Davidson, N. E., and H. H. Hendon, 1989: Downstream development in the Southern Hemisphere monsoon during FGGE/WMONEX. Mon. Wea. Rev., 117, 1458–1470.
Dickinson, M., and J. Molinari, 2002: Mixed Rossby–gravity waves and western Pacific tropical cyclogenesis. Part I: Synoptic evolution. J. Atmos. Sci., 59, 2183–2196.
Evans, J. L., and K. McKinley, 1998: Relative timing of tropical storm lifetime maximum intensity and track recurvature. Meteor. Atmos. Phys., 65, 241–245.
Frank, W. M., 1982: Large-scale characteristics of tropical cyclones. Mon. Wea. Rev., 110, 572–586.
Frank, W. M., and P. E. Roundy, 2006: The role of tropical waves in tropical cyclogenesis. Mon. Wea. Rev., 134, 2397–2417.
Fu, B., T. Li, M. S. Peng, and F. Weng, 2007: Analysis of tropical cyclogenesis in the western North Pacific for 2000 and 2001. Wea. Forecasting, 22, 763–780.
Ge, X., T. Li, and X. Zhou, 2007: Tropical cyclone energy dispersion under vertical shears. Geophys. Res. Lett., 34, L23807, doi:10.1029/2007GL031867.
Hamming, R. W., 1989: Digital Filters. Prentice Hall, 284 pp.
Harr, P. A., 2010: The extratropical transition of tropical cyclones: Structural characteristics, downstream impacts, and forecast challenges. Global Perspectives on Tropical Cyclones: From Science to Mitigation, J. C. L. Chan and J. D. Kepert, Eds., World Scientific, 149–174.
Hartmann, D. L., and E. D. Maloney, 2001: The Madden–Julian oscillation, barotropic dynamics, and North Pacific tropical cyclone formation. Part II: Stochastic barotropic modeling. J. Atmos. Sci., 58, 2559–2570.
Holland, G. J., 1995: Scale interaction in the western Pacific monsoon. Meteor. Atmos. Phys., 56, 57–79.
Hoskins, B. J., I. N. James, and G. N. White, 1983: The shape, propagation, and mean-flow interaction of large-scale weather systems. J. Atmos. Sci., 40, 1595–1612.
Hsu, H.-H., 2005: East Asian monsoon. Intraseasonal Variability in the Atmosphere–Ocean Climate System, K.-M. Lau and D. Waliser, Eds., Springer-Praxis, 63–94.
Hsu, H.-H., and C.-H. Weng, 2001: Northwestward propagation of the intraseasonal oscillation in the western North Pacific during the boreal summer: Structure and mechanism. J. Climate, 14, 3834–3850.
Hsu, H.-H., Y.-L. Chen, A.-K. Lo, C.-H. Hung, W.-S. Kau, and C.-C. Wu, 2008a: Intraseasonal oscillation–tropical cyclone coupling in the western North Pacific during the 2004 typhoon season. Recent Progress in Atmospheric Sciences: Applications to the Asia–Pacific Region, K. N. Liou and M. D. Chou, Eds., World Scientific, 49–65.
Hsu, H.-H., C.-H. Hung, A.-K. Lo, C.-C. Wu, and C.-W. Hung, 2008b: Influence of tropical cyclones on the estimation of climate variability in the tropical western North Pacific. J. Climate, 21, 2960–2975.
Hsu, P.-C., T. Li, and C.-H. Tsou, 2011: Interactions between boreal summer intraseasonal oscillations and synoptic-scale disturbances over the western North Pacific. Part I: Energetics diagnosis. J. Climate, 24, 927–941.
Kaylor, R. E., 1977: Filtering and decimation of digital time series. Tech. Note BN 850, Institute of Physical Science Technology, University of Maryland, College Park, 42 pp.
Ko, K.-C., and H.-H. Hsu, 2006: Sub-monthly circulation features associated with tropical cyclone tracks over the East Asian monsoon area during July–August season. J. Meteor. Soc. Japan, 84, 871–889.
Ko, K.-C., and H.-H. Hsu, 2009: ISO modulation on the submonthly wave pattern and the recurving tropical cyclones in the tropical western North Pacific. J. Climate, 22, 582–599.
Krouse, K. D., A. H. Sobel, and L. M. Polvani, 2008: On the wavelength of the Rossby waves radiated by tropical cyclones. J. Atmos. Sci., 65, 644–654.
Kurihara, Y., M. A. Bender, and R. J. Ross, 1993: An initialization scheme of hurricane models by vortex specification. Mon. Wea. Rev., 121, 2030–2045.
Kurihara, Y., M. A. Bender, R. E. Tuleya, and R. J. Ross, 1995: Improvements in the GFDL hurricane prediction system. Mon. Wea. Rev., 123, 2791–2801.
Lau, K.-H., and N.-C. Lau, 1992: The energetics and propagation dynamics of tropical summertime synoptic-scale disturbances. Mon. Wea. Rev., 120, 2523–2539.
Li, T., 2006: Origin of the summertime synoptic-wave train in the western North Pacific. J. Atmos. Sci., 63, 1093–1102.
Li, T., and B. Fu, 2006: Tropical cyclogenesis associated with Rossby wave energy dispersion of a preexisting typhoon. Part I: Satellite data analyses. J. Atmos. Sci., 63, 1377–1389.
Liebmann, B., H. H. Hendon, and J. D. Glick, 1994: The relationship between tropical cyclones of the western Pacific and Indian Oceans and the Madden–Julian oscillation. J. Meteor. Soc. Japan, 72, 401–412.
Maloney, E. D., and D. L. Hartmann, 2001: The Madden–Julian oscillation, barotropic dynamics, and North Pacific tropical cyclone formation. Part I: Observations. J. Atmos. Sci., 58, 2545–2558.
Maloney, E. D., and M. J. Dickinson, 2003: The intraseasonal oscillation and the energetics of summertime tropical western North Pacific synoptic-scale disturbances. J. Atmos. Sci., 60, 2153–2168.
Shapiro, L. J., and K. V. Ooyama, 1990: Barotropic vortex evolution on a beta plane. J. Atmos. Sci., 47, 170–187.
Sobel, A. H., and C. S. Bretherton, 1999: Development of synoptic-scale disturbances over the summertime tropical northwest Pacific. J. Atmos. Sci., 56, 3106–3127.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 2961–3012.
Wu, C.-C., T.-H. Yen, Y.-H. Kuo, and W. Wang, 2002: Rainfall simulation associated with Typhoon Herb (1996) near Taiwan. Part I: The topographic effect. Wea. Forecasting, 17, 1001–1015.
Yanai, M., S. Esbensen, and J.-H. Chu, 1973: Determination of bulk properties of tropical cloud clusters from large-scale heat and moisture budgets. J. Atmos. Sci., 30, 611–627.
Zhou, C., and T. Li, 2010: Upscale feedback of tropical synoptic variability to intraseasonal oscillations through the nonlinear rectification of the surface latent heat flux. J. Climate, 23, 5738–5754.
