Abstract

This study demonstrates that during the passage of the MJO through the Maritime Continent in the boreal winter, the corresponding deep convection and near-surface wind anomalies tend to skirt around mountainous islands. Flow bifurcation around elongated mountainous islands, such as New Guinea, is clearly seen. Topographic blocking generates distinctive vorticity and convergence distributions in this specific domain. Mountain-wave-like structures are also observed throughout the Maritime Continent, with a clear spatial relationship with the high terrains in Sumatra, Sulawesi, and New Guinea. The existence of topography seems to create extra lifting and sinking within the large-scale circulation and thus the convective system exhibits quasi-stationary features near the major topography during the MJO passage through the Maritime Continent. It is suggested that resolving the detailed topographic effects may play a key role in simulating realistic characteristics of the MJO in the Maritime Continent.

1. Introduction

The Madden–Julian oscillation (MJO; Madden and Julian 1971), a large-scale tropical convection–circulation phenomenon oscillating in the intraseasonal time scale, is well known for its eastward propagation along the equator. This phenomenon is particularly evident during the boreal winter. The MJO exhibits the characteristics of a Kelvin–Rossby wave packet associated with deep convection (Gill 1980). This Gill-type structure in the lower troposphere is characterized by an elongated weaker–easterly (east side)–stronger–westerly (west side) pattern along the equator and two off-equator cyclonic circulations associated with the equatorial westerlies. Several mechanisms, based mostly on the equatorial wave dynamics, have been proposed to explain the eastward propagation tendency (e.g., Lau and Peng 1987; Wang 1988; Salby et al. 1994; Neelin et al. 1987; Emanuel 1987; Lau and Sui 1997; Flatau et al. 1997; Maloney and Hartmann 1998).

The hypotheses mentioned above treat the earth as an aquaplanet and ignore the existence of tropical topography and the land–sea contrast. Recent studies have demonstrated that the topographic effect is likely important in the Maritime Continent where the topography and land–sea contrast are particularly complicated (Fig. 1a). In a study of the MJO during the boreal summer, Hsu et al. (2004) revealed that the MJO does not propagate smoothly eastward, as an equatorial wave does. Instead, it is a combination of stationary and propagating components. As a result, the MJO is characterized by the “sequentially downstream development” of deep convection in the Maritime Continent at specific longitudes, for example, 95°, 110°, 120°, and 145°E, where mountainous islands such as Sumatra, Borneo, Sulawesi, and New Guinea are located. Hsu and Lee (2005) further explored the topographic effect on the MJO in the boreal winter in a global context. Their study suggested that the lifting and frictional effects caused by the topography and land–sea contrast in the Maritime Continent may help induce a near-surface moisture convergence to the east of that topography where a new deep convection region develops. A sudden shift in the deep convection from one region to another is therefore observed when a MJO passes through the Maritime Continent. In a simulation study, Inness and Slingo (2006) suggested that the topographic blocking effect on the Kelvin wave leads to the observed weakening in the MJO. A poor simulation is therefore expected, considering the poor topographical representation in the coarse-resolution model.

Fig. 1.

(a) Terrain height (m) in the Maritime Continent, including part of Indochina, the Philippines, and Australia. (b) Variances of 20–100-day filtered TRMM precipitation (mm2 h−2, shading) and 850-hPa zonal wind (m2 s−2, contour) in the Maritime Continent during the boreal winter.

Fig. 1.

(a) Terrain height (m) in the Maritime Continent, including part of Indochina, the Philippines, and Australia. (b) Variances of 20–100-day filtered TRMM precipitation (mm2 h−2, shading) and 850-hPa zonal wind (m2 s−2, contour) in the Maritime Continent during the boreal winter.

Topography may also significantly influence the smaller-scale features embedded in the MJO. Neale and Slingo (2003) argued that the energy and hydrological cycles of the Maritime Continent are determined by the regional diurnal cycle and complex circulation patterns. It was suggested that the regional modeling approach, which offers finer resolution for capturing diurnal variation and sea breezes, may be able to reduce the simulation difficulty when using the general circulation model. Shibagaki et al. (2006) studied the multiscale aspects of convective systems associated with an intraseasonal oscillation (ISO) event over the Maritime Continent. Their approach decomposes the large-scale ISO into supercloud clusters (SCCs; 2000–4000 km) and mesoscale cloud clusters (MCCs; 500–1000 km), which are in much smaller horizontal scales than ISO. The clusters propagate and develop over the Sumatran mountain range while the precipitation systems evolve from convective to stratiform during this process. Although this study focuses on the cloud-cluster-scale instead of large-scale evolution, it implies how the eastward-propagating MJO is influenced by the fine topography in the Maritime Continent, especially over Sumatra, through multiscale interaction. While most GCMs have difficulty simulating the MJO around the Maritime Continent, Miura et al. (2007) successfully simulated a slow eastward-migrating MJO event using a global cloud-resolving model (GCRM). The success of this 7-km high-resolution GCRM may be attributed to its ability to capture the coupling between atmospheric circulation and clouds, tropical topography, and the SST gradient. Their success suggests that the better-represented topography, land–sea contrast, and the associated cloud evolution processes over the Maritime Continent play a key role in MJO modeling.

While these studies revealed the significance of the topographic effect on the MJO, many detailed structures related to the complex land–sea contrast and topography in the Maritime Continent are not yet fully understood. For example, does the topography induce stationary-wave-like perturbation? Does the large mountain mass such as New Guinea Island block the prevailing flow as an isolated obstacle does? This study, combining both advantages of comprehensive dynamic fields from reanalysis data and fine-resolution convective properties of recent satellite data, extends the previous study by Hsu and Lee (2005) to demonstrate more clearly the geographical relationship between the anomalous circulation–convection of the MJO, the topography, and land–sea contrast in the Maritime Continent. During the boreal summer, the MJO is usually complicated by northward propagation in the northern Indian Ocean, the South China Sea, and the tropical western North Pacific (e.g., Wang and Rui 1990; Hsu et al. 2004) and exhibits weaker eastward propagation through the Maritime Continent. Boreal winter is therefore chosen in the present study to reveal more clearly the topographic effect on the MJO.

A brief introduction of the datasets and MJO composite method are described in section 2. Phase composites of outgoing longwave radiation (OLR)–precipitation properties are shown in section 3. Sections 4 and 5 provide detailed discussion on the relationship between the MJO and the mountainous islands in the Maritime Continent from a dynamical view. Section 6 summarizes the major findings of this study and discusses the implications.

2. Data and methodology

The datasets used in this study include 1) the interpolated OLR (Liebmann and Smith 1996), 2) 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Uppala et al. 2005), and 3) Tropical Rainfall Measuring Mission surface precipitation (TRMM 3B42 version 6; Adler et al. 2000; Kummerow et al. 2000; Iguchi et al. 2000; Haddad et al. 1997a,b). The ERA-40 data cover 13 winters (defined as the period from December to February of the next year) from 1988/89 to 2000/01 on a 2.5° × 2.5° grid, while the TRMM data cover 10 winters from 1998/99 to 2007/08 on a 0.25° × 0.25° grid. The OLR covers a longest period from 1988/89 to 2007/08 on a 2.5° × 2.5° grid.

To extract the intraseasonal signal from the original data, a 20–100-day bandpass Butterworth filter (Kaylor 1977) was applied to all data. A composite technique, following Maloney and Hartmann (1998), is applied to construct the composite MJO cycle. An empirical orthogonal function (EOF) analysis applied to the meridionally (5°N–5°S) averaged OLR yields the first and second EOFs (EOF1 and EOF2), which explain 15.3% and 11.8% of the total variance, respectively, as shown in Fig. 2a. The two leading EOFs display a similar spatial pattern but with a 60° longitudinal phase shift. Figure 2b demonstrates that the second principal component (PC2) leads the first principal component (PC1) by 9 days, as revealed in a lag-correlation calculation. The close temporal and spatial phase relationship indicates that the EOF1/PC1 and EOF2/PC2 together form an eastward-propagating MJO. Therefore, the MJO index is defined as the sum of the PC1 at time t and the PC2 9 days before (t − 9):

 
formula

where t denotes the time in days.

Fig. 2.

(a) The first (sold line) and second (dashed line) EOF of OLR anomalies averaged from 5°N to 5°S. (b) Lag correlation between principal components of the first and second EOF, from −30 days to +30 days.

Fig. 2.

(a) The first (sold line) and second (dashed line) EOF of OLR anomalies averaged from 5°N to 5°S. (b) Lag correlation between principal components of the first and second EOF, from −30 days to +30 days.

The daily time series for the index was calculated for the 1988–2008 period. Dates with the index greater than one standard deviation were chosen as key dates and defined as phase 5, which is characterized by the negative OLR anomaly (i.e., active convection) in the Maritime Continent. After the dates of phase 5 were decided, phases 1 and 9 were defined as the nearest dates when the local minima are less than zero. Phases 3 and 7 are the zero value points between phases 1, 5, and 9. The remaining phases were obtained using an equidistant interpolation between phases 1, 3, 5, 7, and 9. All cases for each parameter were averaged for each phase to yield the composite MJO cycle.

Twenty-two cases from 1988/89 to 2000/01 were chosen for ERA-40 data composite, and 17 cases from 1998/99 to 2007/08 were chosen for TRMM data. Although the case numbers are different in these datasets, the composites of different variables exhibit characteristics that are mutually consistent. As demonstrated later, the difference in case numbers does not affect our results and interpretation of the major characteristics of the MJO in the Maritime Continent. Statistical significance test is then applied to the composites, and the patterns that are significant at the 0.1 level are highlighted in the following figures.

The Japanese 25-yr Reanalysis (JRA-25) (Onogi et al. 2007) and National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis I (Kalnay et al. 1996), as well as Quick Scatterometer (QuikSCAT) surface wind from satellite retrieval (Dunbar et al. 2006), were also used to verify the composite results of the 3D wind structure and 10-m wind retrieved from ERA-40 reanalysis data. The comparison reveals similar patterns among these datasets (not shown). It follows that the topographic effect revealed in the ERA-40 is also present in other reanalysis and datasets.

3. Convection, near-surface circulation, and topography

The Maritime Continent is so called because of the fragmentary distribution of islands and seas (Fig. 1a). Narrow and high mountains on the islands of Sumatra, Borneo, Sulawesi, and New Guinea create sharp terrain variations in this domain. Several of these islands have mountains higher than 3000 m. Long mountain ranges exist on elongated islands, such as Sumatra, Java, and New Guinea. The largest one is the northwest–southeast mountain range, which extends over 1600 km with the highest mountain at 4884 m, on the world’s second largest island, New Guinea. It will be demonstrated later that these narrow and long mountain ranges seem to have a significant effect on the MJO.

Intraseasonal variance accounts for 40%–50% and 15%–30% of total variance in 850-hPa zonal wind and TRMM surface precipitation, respectively. The spatial distribution of intraseasonal variance presented in Fig. 1b exhibits distinct land–sea contrast in the Maritime Continent, especially in the vicinity of larger islands such as Sumatra, Borneo, and New Guinea. In the boreal winter, the zonal wind variance is larger south of the equator than north of the equator. Furthermore, variance over ocean is generally larger than over land. For example, local zonal wind variance maxima reside over the oceanic channel between islands in the Maritime Continent such as Java, New Guinea, and the Australian landmass. The other maximum locates at the northeast side of New Guinea. These maxima distinctively contrast the local minimum variances along a zonal band covering Borneo, Sulawesi, and New Guinea. Another noteworthy feature is the maximum–minimum–maximum distribution from the Torres Strait to the region north of New Guinea, along the 130°–150°E longitudinal bands.

The TRMM precipitation variance exhibits similar land–sea contrast. Large variances are found south of Sumatra and Java, over the Java Sea, the Arafura Sea, the Gulf of Carpentaria, and the western Pacific. Same as for zonal wind, the precipitation variance over the oceanic regions north and south of New Guinea is much larger than the variance over New Guinea. As noted by Hsu and Lee (2005), these variance maxima lay along the propagation path of the MJO through the Maritime Continent. Other maxima are located near the Malay Peninsula, the South China Sea, and the Philippines. These are places where convection activity fluctuates during the MJO passage.

The TRMM surface precipitation (shading) and OLR (contour) anomaly maps between 12.5°N and 12.5°S from phase 1 to phase 7 are shown in Fig. 3. Note that only those statistically significant at the 0.1 level are shown. Although the temporal TRMM and OLR coverage overlaps by only three winters (1998/99–2000/2001), the respective phase evolution and dominant features are quite alike. The consistency between the OLR and TRMM composites indicates that different temporal coverage does not affect our observation about the MJO because the major characteristic of the MJO is rather stationary in time, at least in the past few decades. The close association between positive TRMM precipitation anomaly and negative OLR anomaly suggests that the TRMM precipitation anomaly can be interpreted as the fluctuation of deep convection. In the following discussion the positive precipitation anomaly, which also has finer spatial resolution for revealing details, will be interpreted as active deep convection.

Fig. 3.

TRMM surface precipitation (shading) and OLR anomalies (contour) from phase 1 to phase 7. Shadings are plotted from 0.1 to 0.3 mm h−1 for every 0.1 mm h−1, and contours are from −25 to −15 W m−2 for every 5 W m−2. Only those that are significant at the 0.1 level are shown.

Fig. 3.

TRMM surface precipitation (shading) and OLR anomalies (contour) from phase 1 to phase 7. Shadings are plotted from 0.1 to 0.3 mm h−1 for every 0.1 mm h−1, and contours are from −25 to −15 W m−2 for every 5 W m−2. Only those that are significant at the 0.1 level are shown.

The eastward propagation tendency of the MJO is clearly observed in both TRMM precipitation and OLR anomalies in Fig. 3, but with the apparent influence of the land–sea contrast in the Maritime Continent. During the period from phase 1 to phase 3, the major MJO convection anomaly propagates along the equator from 75° to 100°E. When reaching the northwest–southeast-elongated Sumatra at phase 4, the convection anomaly moves southeastward along the southwest edge of Sumatra to the Timor Sea, while anomalous convection also develops over the Java Sea, the Banda Sea, and over the ocean to the northwest of New Guinea. The relatively free convection over the neighboring islands is evident. At phase 5, a region of anomalous convection flares suddenly over the ocean to the northeast of New Guinea, while the convection is still active over the ocean to the west of New Guinea but is inactive over New Guinea. This sudden development between phase 4 and phase 6 shifts the major convection center from 130° to 150°E and closer to the equator. The newly formed convection anomaly east of New Guinea continues its southeastward journey to the western Pacific. The eastward movement of the MJO convection apparently skirts around the island, explaining the large intraseasonal variance over the ocean. The presence of the island chain from Sumatra to New Guinea in the Maritime Continent seems to have an effect in detouring the movement of the MJO deep convection, first southward when entering the Maritime Continent and then back to near the equator when leaving.

As noted in previous studies (e.g., Ichikawa and Yasunari 2008), diurnal fluctuation is strong over the islands in the Maritime Continent, apparently because of the heating effect of the land surface. Conversely, the mountainous islands tend to prevent the convection from occurring over lands in the intraseasonal time scale. The existence of the high-mountain island seemingly leads to the skirting and jumping of the major convection around the islands. An examination of near-surface circulation may provide more clues to this interesting characteristic. Horizontal maps of TRMM precipitation (contours/shadings) and 10-m wind anomalies (vectors) from phase 2 to phase 7 are shown in Fig. 4.

Fig. 4.

TRMM surface precipitation (contour/shading) and 10-m wind anomalies (m s−1, vector) from phase 2 to phase 7. Contours are plotted from 0.1 to 0.3 mm h−1 for every 0.1 mm h−1. Shadings and vectors in black denote those that are significant at the 0.1 level.

Fig. 4.

TRMM surface precipitation (contour/shading) and 10-m wind anomalies (m s−1, vector) from phase 2 to phase 7. Contours are plotted from 0.1 to 0.3 mm h−1 for every 0.1 mm h−1. Shadings and vectors in black denote those that are significant at the 0.1 level.

During the period when the wind pattern changes from the prevailing easterly anomaly over the entire Maritime Continent at phase 2 to the prevailing westerly anomaly at phase 6, the development of the westerly anomaly occurs mainly over the oceanic regions, and following closely the movement of the deep convection it appears in and also to the west of the deep convection region. The westerly anomaly tends to split toward the north and south when it approaches the mountainous islands and flows around the major terrains. This can be seen clearly to the west of Sumatra at phase 3, to the north and south of Java at phase 4, and to the north and south of Sulawesi and New Guinea at phases 5–7. Strong westerly anomalies are observed both to the north and south of these islands, but very weak winds are observed atop the islands. The splitting flow around these islands does not occur only in the westerly phase. It is also evident at phase 2 when the easterly anomaly prevails in the Maritime Continent. Similar results were also observed in the pressure-level wind anomaly in the lower troposphere, such as 850 hPa. More details will be discussed in the latter sections.

4. Vorticity and divergence

Figure 5 shows vorticity and divergence anomalies at 850 hPa from phase 2 to phase 7, respectively. Before the arrival of the MJO convection center at Sumatra (phase 2), the meridional structure of vorticity pattern between 120° and 150°E is basically characterized by four zonally elongated bands of anomalies: negative to the north of 5°N, positive near the equator (except Borneo), negative between the equator and 10°S, and positive between 10° and 20°S. These zonally banded patterns are distorted around mountainous islands, where the splitting flows are observed in Fig. 4. The most interesting ones are the positive–negative vorticity dipoles over Sulawesi and New Guinea. At phase 2 when the easterly anomaly prevails, positive and negative vorticity anomalies are located in the north and south of these two islands, respectively. Conversely, the vorticity dipole reverses phase and becomes negative in the north and positive in the south at phase 4 and 5 when the prevailing wind is westerly anomaly. The existence of the vorticity dipole can be attributed to the blocking and frictional effect of the mountains on the prevailing flow. Both easterly and westerly anomalies at 850 hPa, the same as the 10-m wind shown in Fig. 4, are very weak near the mountainous islands and increasingly stronger away from the islands (not shown). This shearing effect that will be further discussed in Fig. 5 would result in the existence of the vorticity dipoles over Sulawesi and New Guinea. Such a dipole may also exist around the elongated mountainous Java Island. However, data in a much higher spatial resolution will be needed to reveal the structure because of the narrowness of Java. Another interesting feature is the vorticity anomaly over Borneo. At phase 2, easterly anomaly curves clockwise around southern Borneo (as in Fig. 4a), likely because of the blocking effect of mountainous Borneo, and results in negative vorticity anomaly. On the contrary, positive vorticity anomaly appears at phase 5 when westerly anomaly curves counterclockwise around southern Borneo (e.g., Fig. 4d).

Fig. 5.

Vorticity (contour) and divergence (shading) anomalies at 850 hPa from phase 2 to phase 7. Solid lines are plotted from 1 × 10−6 to 4 × 10−6 s−1 for every 1 × 10−6 s−1, and dashed lines are from −4 × 10−6 to −1 × 10−6 s−1 for every 1 × 10−6 s−1. Shadings are plotted from −2 × 10−6 to 2 × 10−6 s−1 for every 0.5 × 10−6 s−1.

Fig. 5.

Vorticity (contour) and divergence (shading) anomalies at 850 hPa from phase 2 to phase 7. Solid lines are plotted from 1 × 10−6 to 4 × 10−6 s−1 for every 1 × 10−6 s−1, and dashed lines are from −4 × 10−6 to −1 × 10−6 s−1 for every 1 × 10−6 s−1. Shadings are plotted from −2 × 10−6 to 2 × 10−6 s−1 for every 0.5 × 10−6 s−1.

Generally speaking, a MJO is characterized by the Kelvin–Rossby wave packet, with twin cyclonic circulation and equatorial westerly to the west of deep convection and equatorial easterly to the east. Such a large-scale pattern is distorted when the MJO moves into the Maritime Continent because of the existence of the mountainous islands. The distortion of the easterly anomaly can be seen clearly in Fig. 4a. To the east of New Guinea (e.g., 150°E) where few islands exist, the easterly anomaly is almost uniformly distributed and weakens away from the equator, exhibiting the Kelvin wave characteristics. Conversely, the easterly anomaly in the Maritime Continent prevails only over the oceanic channels between islands. This land–sea contrast is also observed in the vorticity field. For example, the positive vorticity anomaly associated with the anticyclonic circulation anomaly south of the equator at phase 2 is meridionally wider over the region from the Indian Ocean to the west of Australia, but it becomes much narrower in the Maritime Continent and restricted in the oceanic channel between the islands and Australia where a stronger easterly anomaly is observed. This land–sea contrast effect results in the distorted vorticity distribution, whose shape more or less follows the land–sea distribution. This characteristic is also evident in the later phases (e.g., phases 4–6) when the westerly and negative vorticity anomalies prevail. The circulation at phases 5 and 6 is in the equatorial Rossby wave regime, which should be characterized by positive and negative vorticity to the north and south of the equator, respectively. However, this vorticity pair is interrupted by the vorticity dipoles existing over Sulawesi and New Guinea and becomes a quadruple pattern in the Maritime Continent as discussed before.

The divergence anomalies presented in Fig. 5 show good relationships with the surface precipitation in Fig. 4, where convergence zones coincide with large precipitation regions. When the MJO convection anomaly moves closer to the west of Sumatra at phase 2, anomalous convergence is observed at 90°–100°E and anomalous divergence at 140°–160°E, implying the horizontal scale of the initial vertical circulation is about 70° longitude. Embedded in this large-scale convergence–divergence dipole are local convergence and divergence pairs near mountainous islands. For example, convergence anomalies are found near Sulawesi and at the western end of New Guinea Island (e.g., 120° and 135°E near the equator), while a divergence anomaly is found at the east side of New Guinea Island (e.g., 140°E). This local east–west convergence–divergence pair is associated with the north–south vorticity dipoles over Sulawesi and New Guinea. It can be understood as follows: the topographic blocking effect on the easterly flow leads to local divergence and convergence anomaly at the windward (eastern) and lee (western) side of topography, respectively. The speed of splitting easterly anomaly around the topography decreases toward the mountain terrain and creates local counterclockwise–clockwise circulation to form the north–south vorticity dipoles. After phase 2, the convergence anomaly to the west of Sumatra moves southeastward along the southwestern edge of Sumatra and Java. This southward shift of convergence anomalies, which is consistent with the movement of the precipitation anomaly shown in Figs. 3 and 4, is likely due to the blocking effect of the elongated mountainous Sumatra. By phase 5 when the convergence anomaly moves to the west and south of New Guinea, the convergence anomaly on the east side of New Guinea becomes well developed and merges with the southern pattern between 140° and 160°E. At this time, the large-scale convergence–divergence dipole switches sign, with anomalous convergence and divergence at the eastern and western sides of the Maritime Continent, respectively. The divergence–convergence pairs over Sulawesi and New Guinea also change signs with the divergence and convergence anomalies at the western and eastern ends, respectively. The vorticity dipoles straddling the islands reverse their signs at the same time. The evolution of the low-level divergence–convergence anomaly is also strongly influenced by the topography like other variables.

5. Cross sections of dynamical fields

a. Meridional profiles at specific longitudes

The topographic effect on the MJO circulation and convection is demonstrated further in this section by examining the meridional vertical cross section at longitudes where mountainous islands are located. Zonal wind (shadings), vorticity (contours), and circulation (vectors) anomalies at 100°E during phase 4 (Fig. 6a), at 120° and 140°E during phase 6 (Figs. 6b,c), and at 160°E (Fig. 6d) during phase 7 are shown. These specific phases are chosen when a strong westerly anomaly, instead of convection, prevails near the mountainous islands to clearly reveal the topographic effect on the flow. At phase 4 when strong westerly anomalies reach Java and a cyclonic circulation forms to the south of 10°S (Fig. 4c) in the southeastern Indian Ocean, the cross section at 100°E (Fig. 6a) is characterized by a pair of deep vorticity anomalies, positive in the Northern Hemisphere and negative in the South Hemisphere. This vorticity anomaly pair may be viewed as the Rossby wave located to the west of deep convection, which is now located south of Sumatra and Java. The asymmetry between the two vorticity components is apparent. For example, the southern negative one exhibits a deep vertical structure with maximum amplitudes in the lower troposphere, while the northern positive one is an elevated deep structure with maximum amplitudes in the middle troposphere. Furthermore, a shallow (below 850 hPa) positive vorticity anomaly is found between the southern negative vorticity anomaly and the mountainous Sumatra (dark-shaded area in the figure). At the same time, a westerly anomaly with the maximum below 600 hPa is found between the positive–negative vorticity pair south of Sumatra. The wind speed anomaly decreases quickly away from the center. The strong zonal wind shear at both sides of the maximum wind anomaly apparently results in the near-surface vorticity anomaly. In comparison with the flow pattern at 120° and 140°E, which will be discussed later, the shifting of the zonal wind anomaly at 100°E to the Southern Hemisphere is likely due to the blocking effect of the elongated mountainous Sumatra, which also results in the near-surface positive anomaly between 5°S and the equator.

Fig. 6.

Cross sections of zonal wind anomalies (m s−1, shading), vorticity (contour), and circulation (vector) anomalies at different longitudes: (a) phase 4 at 100°E, (b) phase 6 at 120°E, (c) phase 6 at 140°E, and (d) phase 7 at 160°E. Shadings are plotted from 1 to 4 m s−1 for every 1 m s−1. Contour interval for vorticity is 1 × 10−6 s−1, and positive negative contours are plotted in solid and dashed lines, respectively. The black bars mark terrain heights. Vertical motion has been multiplied by 100. Shadings and vectors in black denote those that are significant at the 0.1 level.

Fig. 6.

Cross sections of zonal wind anomalies (m s−1, shading), vorticity (contour), and circulation (vector) anomalies at different longitudes: (a) phase 4 at 100°E, (b) phase 6 at 120°E, (c) phase 6 at 140°E, and (d) phase 7 at 160°E. Shadings are plotted from 1 to 4 m s−1 for every 1 m s−1. Contour interval for vorticity is 1 × 10−6 s−1, and positive negative contours are plotted in solid and dashed lines, respectively. The black bars mark terrain heights. Vertical motion has been multiplied by 100. Shadings and vectors in black denote those that are significant at the 0.1 level.

Accompanying the vorticity pair and the westerly wind anomaly in the Southern Hemisphere is a vertical circulation with the descent over and to the south of Sumatra and the ascent around 10°S. The ascent corresponds to the deep convection embedded in the MJO (Figs. 4b,c). The descending motion is apparently due to the blocking effect of the elongated mountainous Sumatra, which results in the bifurcation of the westerly flow to the north and south (Figs. 4b,c) and the divergence in the lower troposphere (Figs. 5b,c). It is interesting to note that the near-surface northerly anomaly in this vertical circulation may act to accelerate the westerly anomaly through the Coriolis force.

The situation is different at 120°E where the topographic effect is even more interesting. Topography in this area includes Sulawesi (the equator to 5°S), Sumba of the Lesser Sunda Islands (10°S), and northern Australia (south of 20°S). Shown in Fig. 6b is the cross section at phase 6 when the topographic effect is most evident. When the westerly anomalies start prevailing at phase 5, the westerly anomaly flows over the mountainous Sulawesi. The overall distribution of vorticity anomaly is negative in the Southern Hemisphere and positive in the Northern Hemisphere. The distortion of this vorticity pair by the topography is obvious. While the westerly anomaly flows over the mountains, especially over Sulawesi, its speed decreases toward the mountains both meridionally and vertically and results in an archlike distribution over Sulawesi and Sumba of the Lesser Sunda Islands. As a result, shallow negative and positive vorticity anomalies appear over northern and southern Sulawesi, respectively. A comparison between total vorticity (∂υ/∂x – ∂u/∂y) and shear vorticity (−∂u/∂y) confirms that the vorticity dipole owes its existence to the strong zonal wind shear near the topography. This result indicates again the importance of the topography in modifying the MJO structure. A similar tendency is also seen over Sumba of the Lesser Sunda Islands.

The effect of the mountain ranges in New Guinea on the MJO is demonstrated below. When the MJO reaches the central Maritime Continent at phase 5, the westerly anomaly around New Guinea splits into two branches, as seen in Fig. 4d. This phenomenon can be seen clearly at phase 6, shown in the vertical cross section at 140°E (Fig. 6c). A westerly anomaly extending from the surface to 400 hPa exists both north and south of New Guinea. The maximum wind speed is observed over the equator and around 10°S in the low–middle troposphere and decreases meridionally away from the maximum wind region, leaving a weak westerly wind anomaly atop New Guinea. The large zonal wind shear at both sides of the mountain results in the negative and positive vorticity anomaly to the north and south of New Guinea, respectively. The vorticity anomaly exhibits essentially an equivalent barotropic vertical structure. To the north and south of this vorticity dipole around New Guinea, there exists another pair of vorticity anomalies (positive in the north and negative in the south) just like in other cross-section plots. The overall vorticity distribution is characterized by a quadruple pattern, which is seen also in Figs. 5d,e. The presence of mountainous New Guinea apparently results in the appearance of the quadruple vorticity pattern, which is a deep structure extending from surface to the upper troposphere. Anomalous ascent is found to the north of New Guinea and between 10° and 20°S (i.e., the Gulf of Carpentaria in northern Australia), corresponding to the positive precipitation anomaly, while descent is observed over New Guinea and immediately south of it. A meridional vertical circulation is clearly seen to the south of New Guinea and the corresponding near-surface northerly may have an effect in accelerating the westerly wind anomaly through the Coriolis force.

To contrast the topographic effect presented above, the cross section at 160°E, where only a small mountainous island exists and therefore the topographic effect is minimum, is also presented in Fig. 6d. The westerly anomaly is found in the lower–middle troposphere mostly south of the equator. This tendency for the westerly anomaly to appear in the Southern Hemisphere in the western Pacific is one of the major characteristics of the MJO in the boreal winter. The maximum wind speed region is much wider than those in the previous three cross sections where large topography exists. The westerly anomaly region is sandwiched between a pair of vorticity anomaly associated with the cyclonic circulation off the equator. This pattern exhibits the typical characteristics of the circulation, which is located to the west of the deep convection in a MJO. In contrast, the typical cyclonic pair in the MJO is either distorted or modified by the topography in the previous three cross sections.

b. Zonal profiles of 5° and 10°S

Figures 7a–d show the cross sections of anomalous vertical circulation along 5°S, from phase 3 to phase 6. They demonstrate the relationship between the vertical circulation and the topography along the major path of the MJO. The high terrains hindering the passage of the MJO through the Maritime Continent are Sumatra (105°E), Sulawesi (120°E), and New Guinea (140°–150°E). Note that the vertical velocity has been multiplied by 100 to show more clearly the corresponding fluctuation. At phase 3, the major upward and downward motion is located to the west of 120°E and east of 150°E, respectively. In an idealized situation (e.g., an aquaplanet or a homogeneous background state), the east–west overturning circulation would look like a smooth Walker circulation, without much interruption between the upward branch around 90°–100°E and the downward branch east of 150°E. However, wavelike perturbations are observed in the vicinity of the major terrain. The ascending region to the west of 120°E, which corresponds to the major convection region in the MJO in Fig. 4b, is interrupted by the subsidence (below 700 hPa) at the windward side of Sumatra. This subsidence is consistent with the splitting flow and the near-surface divergence in the region due to the blocking effect of Sumatra on the prevailing westerly anomaly in the lower troposphere.

Fig. 7.

Cross sections of vertical velocity anomalies (shading) and streamlines (solid line) at different latitudes: (a)–(d) phase 3 to phase 6 at 5°S and (e)–(h) phase 3 to phase 6 at 10°S. The shadings are plotted from −0.01 to −0.04 hPa s−1 every 0.01 hPa s−1. The black bars mark terrain heights. Vertical motion has been multiplied by 100, and shading indicates anomalously upward motion. Shadings denote those that are significant at the 0.1 level.

Fig. 7.

Cross sections of vertical velocity anomalies (shading) and streamlines (solid line) at different latitudes: (a)–(d) phase 3 to phase 6 at 5°S and (e)–(h) phase 3 to phase 6 at 10°S. The shadings are plotted from −0.01 to −0.04 hPa s−1 every 0.01 hPa s−1. The black bars mark terrain heights. Vertical motion has been multiplied by 100, and shading indicates anomalously upward motion. Shadings denote those that are significant at the 0.1 level.

In the next phase, the major ascending region shifts eastward to 135°E, corresponding to the eastward propagation of deep convection over the Java Sea (Fig. 4c), and occupies the western half of the domain. Interestingly, another ascending region is found to the east of the high-rising mountain in New Guinea, located between 140° and 165°E, where the easterly anomaly prevails. At this stage, the whole domain is dominated by anomalously upward motion and deep convection, except the descent near major mountains, especially the narrow descending region near 120°E and immediately to the west of New Guinea between 135° and 140°E. The phase 5 is characterized by two counterclockwise overturning circulations, with the ascending and descending at the eastern and western ends, respectively. The western overturning circulation is already observed at phase 4, but now with a smaller zonal scale, because of the moving in of the descending branch at the western boundary of the domain. The second overturning circulation is a new one, which is developed atop and to the east of New Guinea. The appearance of this vertical circulation corresponds to the sudden development of the deep convection to the northeast of New Guinea at phase 5 seen in Fig. 4d.

Generally speaking, the zonal circulation in the domain is characterized by wavelike disturbances, with alternating ascending and descending anomalies at the windward and lee sides of mountainous islands between 100° and 160°E. An exception is found in the region around New Guinea, where descent and ascent are observed at the windward and lee sides of New Guinea, respectively. The descent has been there since phase 4, but became stronger and more organized at phase 5. Different flow characteristics may be caused by different mechanisms. The mountain range in New Guinea is much larger and higher than the mountains in other islands. The blocking effect of New Guinea is likely to be more significant. For example, the descending region is where the bifurcation of 10-m wind and near-surface divergence occurs, as seen in Fig. 4d. The meridional profiles of zonal wind speed at 140°E indicate that the flow bifurcation is not just a near-surface phenomenon; instead, it extends all the way to the upper troposphere (Fig. 6c). The blocking effect of the high-rising mountain in New Guinea may force the low-level wind to split and flow around the elongated mountain range. It may in turn induce subsidence to the west of the island, where the near-surface divergence occurs (Fig. 5d). On the contrary, the ascending region occurs to the east of New Guinea, where flow converges. This distinct characteristic is also evident at phase 6 (Fig. 7d).

The major difference between phases 4 and 5 is the weakening of the deep convection in the western Maritime Continent and the sudden appearance of the deep convection to the east of New Guinea. In the next phase, major circulation features observed at phase 5 remain evident, while the convection to the west of New Guinea starts weakening. A careful examination of the circulation from phase 3 to phase 5 reveals that strong ascending and descending anomalies tend to occur at the same regions, for example, the ascending regions west of Sumatra (100°E), west of Sulawesi (120°E), over the Java Sea where a strong convergence occurs, and east of New Guinea. Overall, the ascending and descending motion tends to occur at the windward and lee sides of the mountains in Sumatra and Sulawesi, but descent and ascent occur at the windward and lee sides of New Guinea, respectively. (This difference between Sumatra–Sulawesi and New Guinea will be discussed later.) Therefore, an ascending region in the westerly anomaly regime may become a descending region in the easterly anomaly regime. The wavelike structures tend to occur in specific regions and show little propagation tendency. This suggests the quasi-stationary nature of the wavelike structures, which owe their existence to the complex terrains in the region because of the lifting and blocking effects of the mountains.

While the vertical circulation along 5°S is strongly affected by the mountains in the region, the vertical circulation along 10°S, which is mostly over the narrow oceanic channel between the islands in the Maritime Continent and the Australian mainland, is much less affected by the topography. Compared with its counterpart along 5°S, the cross section along 10°S presented in Figs. 7e–h exhibits the characteristics of a much smoother Walker-type circulation with a larger zonal scale. At phase 2, an ascending motion occurs at the western end of the domain, while a descending motion occurs at the eastern end, indicating a zonal scale of about 80° longitude. Contrary to the combination of stationary and eastward-propagating features along 5°S, the vertical circulation along 10°S moves at almost a constant speed from the western to eastern Maritime Continent from phase 2 to phase 6. The tendency for the vertical motion to occur in certain regions is much less evident. The contrast between cross sections at 5° and 10°S demonstrate how significantly the MJO circulation and convection can be influenced by topography. The MJO is usually explained in terms of large-scale tropical wave (e.g., wavenumber 1–2) and is often treated as global wave pattern in some diagnostic studies. The results shown above indicate that the MJO in the Maritime Continent is significantly modified by local land–sea contrast and topography.

One may wonder whether a spatial smoothing will remove these local effects and yield a large-scale pattern, as documented in many previous studies. To remove these small-scale features, 9-point zonal averaging was applied twice to the parameters shown in Fig. 7, including the topography. The results are shown in Fig. 8. The topography along 5° and 10°S is indeed smoothed out via this method, and the vertical circulation patterns look like a smooth Walker circulation without interruption as seen in Fig. 7. Figures 8e–h show the smooth propagation of the Walker circulation accompanied with the MJO convection at 10°S, which is almost as idealized as may be expected from theoretical studies. The upward branch of the circulation moves into the domain at phase 3, and the whole structure of the overturning circulation shows up in the Maritime Continent at phase 6 with a zonal scale of about 80° longitude. In contrast, the circulation deformation due to the presence of topography at 5°S is still evident (Figs. 8a–d). A saddle point at 800 hPa is found near 140°E at phase 4, which is a separation point for the easterly and westerly anomalies and also the upward and downward motion. The presence of the high-rising mountain range in New Guinea is apparently the cause for the appearance of the saddle point. The curvature bending upward in the lower part of the Walker circulation at phase 6 is obviously the footprint of the topography-induced wave structures. Similar results were obtained by retaining only wavenumber 1–3 components. These results imply that the topographic effect induced by complex topography and land–sea contrast is hard to remove via spatial smoothing or retaining only the longest spatial Fourier components. An analysis based on spatially smoothed data may still retain the influence of local topographic effect. This spatial smoothing or filtering approach has been a common practice in many studies (e.g., Hendon and Salby 1994) to interpret the intraseasonal signals in terms of idealized equatorial waves (Matsuno 1966) in an aquaplanet. The results presented here suggest the potential limitation of this type of approach in comparing the real world with the idealized wave pattern.

Fig. 8.

As in Fig. 7, but all variables including topography are zonally smoothed twice by 9-point averaging.

Fig. 8.

As in Fig. 7, but all variables including topography are zonally smoothed twice by 9-point averaging.

6. Summary and discussion

This study explores the potential effect of the topography on the propagation and characteristics of the MJO in the Maritime Continent. It is demonstrated that the passage of the MJO through the Maritime Continent is not a smooth propagation, as expected in an idealized situation (e.g., an aquaplanet). The mountainous islands exert “blocking effect” on the MJO. As a result, the eastward movement of deep convection and near-surface wind anomalies in the MJO skirt around islands. Because of these effects on the associated vorticity/convergence evolution, the deep convection and westerly anomaly shift southward of the equator, instead of along the equator as in the Indian Ocean, and propagate eastward over the Java Sea, the oceanic region off the southern coast of Sumatra and Java, and the Timor Sea. The propagation is further stalled west of mountainous New Guinea. The westerly anomaly splits and flows around the island and converges at the oceanic region to the northeast of New Guinea. This newly developed convection region, now located near the equator, becomes the major deep convection region of the MJO and moves eastward. The distribution of mountainous islands in the Maritime Continent seems to result in the southward detour of the eastward-propagating MJO and the sudden shift of deep convection from one region to another. In addition to the blocking effects on the low-level wind, mountain-wave-like structures are also observed in the specific longitudes near the high terrains of Sumatra, Sulawesi, and New Guinea. The existence of topography seems to create extra lifting and sinking within the large-scale circulation, and thus the convective systems are observed to diminish or generate on different sides of the tropical topography.

Compared with the theoretical view of the frictional wave-conditional instability of the second kind (CISK) mechanism, which is often applied to explain the MJO, the topographic effect is likely to play an important role in determining the location of deep convection and the flow distribution in finer scales. The topographic effect creates quasi-stationary features and breaks up the larger-scale convection in places. The end result is the stalling propagation and weakening strength of the MJO after it reaches the Maritime Continent. The flow bifurcation around New Guinea and Sulawesi is particularly evident. This phenomenon bears certain similarities to the theoretical characteristics of the flow around an idealized obstacle in the low Froude number condition (Hunt and Snyder 1980; Smolarkiewicz and Rotunno 1989; Sha et al. 1998). The major discrepancy is as follows. In the theoretical study, a pair of lee vortices occurs behind the obstacle. In this study the pair of vorticity anomalies is found straddling the mountain ranges because of the decreasing wind speed toward the mountains. It is hypothesized that frictional effect weakens the wind speed and creates anomalous shear vorticity at the both sides of the mountains. The bifurcation of the flow in the windward side also creates anomalously downward motion, while the flow convergence in the lee side creates anomalously upward motion.

Our results suggest that the topographic effects and the mountain-wave-like perturbations may have a combined effect on the MJO, with different degrees of influence at different regions. It is difficult to quantitatively assess the relative contribution of the blocking and wave-making effect by the topography in the present diagnostic study. The topographic effect on the flow is very complicated and is significantly different from island to island. Well-designed numerical studies are needed to fully understand the topographic effect of each mountainous island. This study only provides qualitative discussion as follows: overall, more spatially elongated and higher mountainous islands exert stronger blocking effect on the incoming flow. The blocking effect of the elongated and high-rising New Guinea is so strong that it causes a complete flow bifurcation from the surface to above 500 hPa. The long island chain of Sumatra and Java not only causes the flow bifurcation but also results in the southward deflection of the incoming westerly anomaly. On the other hand, the lower terrain of Sumatra allows the westerly anomaly to flow over and creates vertical wavelike perturbation in the downstream. While less spatially extended islands such as Sulawesi and Borneo exert only a localized blocking effect, they seem to cause significant downstream wavelike perturbation in the vertical. Flow bifurcation induces local convergence–divergence around the island. The topographically induced vertical wavelike perturbation also induces ascending–descending branches near and downstream of the islands. Both blocking and wave-making effects induce extra enhancement–suppression to the MJO convection activity and therefore modify the behavior of the MJO in the Maritime Continent.

The relative importance of blocking and wave-making effects seems to result in different anomalous ascent–descent distribution near the mountainous islands. In the prevailing anomalous westerly, the strong blocking effect of the elongated and high-rising New Guinea results in local divergence and convergence anomalies near the surface at the western and eastern ends of New Guinea, which are responsible for the local subsidence and ascent, respectively. In the case of Sumatra and Sulawesi, the blocking effect is much weaker, partly because of the lower terrain. The westerly anomaly that is only partially blocked is able to flow over the terrain and induces ascent and descent in the windward and lee side of the mountains.

Further theoretical and numerical studies are required to understand the exact physical mechanism. The results presented above imply that a high-resolution model, which can resolve the detailed topographic effects, may be required to simulate the realistic characteristics of the MJO in the Maritime Continent, even though high resolution does not necessarily guarantee successful simulation, as demonstrated by Rajendran et al. (2008).

Acknowledgments

This study was supported by the National Science Council in Taiwan under Grant NSC-95-2111-M-002-010-MY3. Authors appreciate the precious comments from the three anonymous reviewers.

REFERENCES

REFERENCES
Adler
,
R. F.
,
G. J.
Huffman
,
D. T.
Bolvin
,
S.
Curtis
, and
E. J.
Nelkin
,
2000
:
Tropical rainfall distributions determined using TRMM combined with other satellite and rain gauge information.
J. Appl. Meteor.
,
39
,
2007
2023
.
Dunbar
,
R. S.
, and
Coauthors
,
2006
:
QuikSCAT Science Data Product User’s Manual, Version 3.0.
JPL Document D-18053–Rev A, Jet Propulsion Laboratory, Pasadena, CA, 85 pp
.
Emanuel
,
K. A.
,
1987
:
An air–sea interaction model of intraseasonal oscillations in the tropics.
J. Atmos. Sci.
,
44
,
2324
2340
.
Flatau
,
M.
,
P. J.
Flatau
,
P.
Phoebus
, and
P. P.
Niiler
,
1997
:
The feedback between equatorial convection and local radiative and evaporative processes: The implications for intraseasonal oscillations.
J. Atmos. Sci.
,
54
,
2373
2386
.
Gill
,
A. E.
,
1980
:
Some simple solutions for heat-induced tropical circulation.
Quart. J. Roy. Meteor. Soc.
,
106
,
447
462
.
Haddad
,
Z. S.
,
D. A.
Short
,
S. L.
Durden
,
E.
Im
,
S.
Hensley
,
M. B.
Grable
, and
R. A.
Black
,
1997a
:
A new parameterization of the rain drop size distribution.
IEEE Trans. Geosci. Remote Sens.
,
35
,
532
539
.
Haddad
,
Z. S.
,
E. A.
Smith
,
C. D.
Kummerow
,
T.
Iguchi
,
M. R.
Farrar
,
S. L.
Durden
,
M.
Alves
, and
W. S.
Olson
,
1997b
:
The TRMM ‘Day-1’ radar/radiometer combined rain-profiling algorithm.
J. Meteor. Soc. Japan
,
75
,
799
809
.
Hendon
,
H. H.
, and
M. L.
Salby
,
1994
:
The life cycle of the Madden–Julian oscillation.
J. Atmos. Sci.
,
51
,
2225
2237
.
Hsu
,
H-H.
, and
M. Y.
Lee
,
2005
:
Topographic effects on the eastward propagation and initiation of the Madden–Julian oscillation.
J. Climate
,
18
,
795
809
.
Hsu
,
H-H.
,
C-H.
Weng
, and
C-H.
Wu
,
2004
:
Contrasting characteristics between the northward and eastward propagation of the intraseasonal oscillation during the boreal summer.
J. Climate
,
17
,
727
743
.
Hunt
,
C. R.
, and
W. H.
Snyder
,
1980
:
Experiments on stably and neutrally stratified flow over a model three-dimensional hill.
J. Fluid Mech.
,
96
,
671
704
.
Ichikawa
,
H.
, and
T.
Yasunari
,
2008
:
Intraseasonal variability in diurnal rainfall over New Guinea and the surrounding oceans during austral summer.
J. Climate
,
21
,
2852
2868
.
Iguchi
,
T.
,
T.
Kozu
,
R.
Meneghine
,
J.
Awaka
, and
K.
Okamoto
,
2000
:
Rain-profiling algorithm for the TRMM precipitation radar.
J. Appl. Meteor.
,
39
,
2038
2052
.
Inness
,
P. M.
, and
J. M.
Slingo
,
2006
:
The interaction of the Madden–Julian oscillation with the Maritime Continent in a GCM.
Quart. J. Roy. Meteor. Soc.
,
132
,
1645
1667
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
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
.
Kummerow
,
C.
, and
Coauthors
,
2000
:
The status of the Tropical Rainfall Measuring Mission (TRMM) after two years in orbit.
J. Appl. Meteor.
,
39
,
1965
1982
.
Lau
,
K-M.
, and
L.
Peng
,
1987
:
Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory.
J. Atmos. Sci.
,
44
,
950
972
.
Lau
,
K-M.
, and
C.
Sui
,
1997
:
Mechanism for short-term sea surface temperature regulation: Observations during TOGA COARE.
J. Climate
,
10
,
465
472
.
Liebmann
,
B.
, and
C. A.
Smith
,
1996
:
Description of a complete (interpolated) outgoing longwave radiation dataset.
Bull. Amer. Meteor. Soc.
,
77
,
1275
1277
.
Madden
,
R. A.
, and
P. R.
Julian
,
1971
:
Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific.
J. Atmos. Sci.
,
28
,
702
708
.
Maloney
,
E. D.
, and
D. L.
Hartmann
,
1998
:
Frictional moisture convergence in a composite life cycle of the Madden–Julian oscillation.
J. Climate
,
11
,
2387
2403
.
Matsuno
,
T.
,
1966
:
Quasi-geostrophic motions in the equatorial area.
J. Meteor. Soc. Japan
,
44
,
25
43
.
Miura
,
H.
,
M.
Satoh
,
T.
Nasuno
,
A. T.
Noda
, and
K.
Oouchi
,
2007
:
A Madden-Julian Oscillation event realistically simulated by a global cloud-resolving model.
Science
,
318
,
1763
1765
.
Neale
,
R.
, and
J.
Slingo
,
2003
:
The Maritime Continent and its role in the global climate: A GCM study.
J. Climate
,
16
,
834
848
.
Neelin
,
J. D.
,
I. M.
Held
, and
K. H.
Cook
,
1987
:
Evaporation–wind feedback and low-frequency variability in the tropical atmosphere.
J. Atmos. Sci.
,
44
,
2341
2348
.
Onogi
,
K.
, and
Coauthors
,
2007
:
The JRA-25 Reanalysis.
J. Meteor. Soc. Japan
,
85
,
369
432
.
Rajendran
,
K.
,
A.
Kitoh
,
R.
Mizuta
,
S.
Sajani
, and
T.
Nakazawa
,
2008
:
High-resolution simulation of mean convection and its intraseasonal variability over the tropics in the MRI/JMA 20-km mesh AGCM.
J. Climate
,
21
,
3722
3739
.
Salby
,
M. L.
,
R. R.
Garcia
, and
H. H.
Hendon
,
1994
:
Planetary-scale circulations in the presence of climatological and wave-induced heating.
J. Atmos. Sci.
,
51
,
2344
2367
.
Sha
,
W.
,
K.
Nakabayashi
, and
H.
Ueda
,
1998
:
Numerical study on flow pass of a three-dimensional obstacle under a strong stratification condition.
J. Appl. Meteor.
,
37
,
1047
1054
.
Shibagaki
,
Y.
, and
Coauthors
,
2006
:
Multiscale aspects of convective systems associated with an intraseasonal oscillation over the Indonesian Maritime Continent.
Mon. Wea. Rev.
,
134
,
1682
1696
.
Smolarkiewicz
,
P. K.
, and
R.
Rotunno
,
1989
:
Low Froude number flow past three-dimensional obstacles. Part I: Baroclinically generated lee vortices.
J. Atmos. Sci.
,
46
,
1154
1164
.
Uppala
,
S. M.
, and
Coauthors
,
2005
:
The ERA-40 re-analysis.
Quart. J. Roy. Meteor. Soc.
,
131
,
2961
3012
.
Wang
,
B.
,
1988
:
Dynamics of tropical low-frequency waves: An analysis of the moist Kelvin waves.
J. Atmos. Sci.
,
45
,
2051
2065
.
Wang
,
B.
, and
H.
Rui
,
1990
:
Synoptic climatology of transient tropical intraseasonal convection anomalies: 1975–1985.
Meteor. Atmos. Phys.
,
44
,
43
61
.

Footnotes

Corresponding author address: Dr. Huang-Hsiung Hsu, Department of Atmospheric Sciences, National Taiwan University, Taipei, 10617 Taiwan. Email: hsu@atmos1.as.ntu.edu.tw