Abstract

The observed outgoing longwave radiation (OLR) and ERA-Interim data during 1979–2008 (from November to April) were analyzed to reveal fundamental differences between eastward-propagating (EP) and nonpropagating (NP) MJO events across the Maritime Continent (MC). It was found that when the maximum MJO convection arrives near 120°E, a positive moisture tendency lies in a longitudinal zone (10°S–10°N, 130°–170°E) for the EP cases, whereas a negative tendency appears in the same region for the NP cases. In the latter cases, there are clearly detectable westward-propagating Rossby wave–type dry signals over the equatorial central-western Pacific. The dry Rossby-wave signal may hinder the development of new convection to the east of the MJO convective center, preventing the eastward propagation of the MJO. A moisture budget analysis shows that the positive tendency of specific humidity in the EP composite is mainly attributed to the advection of the mean moisture by an intraseasonal ascending motion anomaly, whereas the negative tendency in the NP composite arises from the advection of anomalous dry air by the mean easterly and the advection of the mean moisture by the anomalous easterly. The EP cases were further separated into two groups: a group with, and a group without, a clear suppressed convective phase of OLR to the east of the MJO convection. In the former (latter), the column-integrated moisture anomaly is negative (positive) to the east of the convection. Nevertheless, MJO crosses the MC in both of the groups, in which anomalous moisture tendency is always positive to the east of the MJO convection. Such positive tendencies are caused by different processes. In the former, anomalous horizontal advections associated with eddy moisture transport and mean moisture advection by intraseasonal meridional wind play an important role. In the latter, it is mainly attributed to mean moisture advection by anomalous vertical velocity.

1. Introduction

As a dominant low-frequency mode in tropical atmosphere, the Madden–Julian oscillation (MJO) exerts a great impact on climate and extreme weather events in various regions of the world (e.g., Madden and Julian 1972; Krishnamurti and Subrahmanyam 1982; Krishnamurti 1985; Murakami and Nakazawa 1985; Li and Wang 2005; Ding and Wang 2007). The MJO signal was first identified by Madden and Julian (1971) in observed single-station data. With more and more available observational data, it became clear that MJO is characterized by the eastward propagation of a large-scale envelope of deep convection along the equator (Weickmann 1983; Weickmann et al. 1985; Lau and Chan 1986), an average phase speed of 4–8 m s−1 over the Indian Ocean and western Pacific warm pool (Wheeler and Kiladis 1999), a planetary zonal scale, and a wide 20–100-day band period (Knutson and Weickmann 1987; Chen and Murakami 1988; Chen et al. 1988; Hartmann and Michelsen 1989; Sperber et al. 1997; Matthews 2008). A number of recent theoretical and observational studies have advanced our understanding of the fundamental dynamics of the MJO [see a review article on this topic by Li (2014)], including its planetary zonal scale selection (Li and Zhou 2009), its multiscale interaction nature (Nakazawa 1988; Majda and Stechmann 2009; Hsu and Li 2011; Wang and Liu 2011), initiation over the western Indian Ocean (IO) (Zhao et al. 2013; Li et al. 2015), and the role of boundary layer zonal moisture asymmetry in eastward propagation (Hsu and Li 2012; Hsu et al. 2014).

While the studies above advanced our understanding of the MJO, current state-of-the-art climate models still have difficulty simulating and predicting MJOs (e.g., Kim et al. 2009). For example, the NCEP CFSv2 model often underestimates the MJO variance in the Maritime Continent (MC) and eastward phase propagation speed (Fu et al. 2013). For observed MJO events that crossed the MC, climate models frequently underestimate the MJO amplitude in the MC; as a result, the model MJO damps quickly before crossing the MC. This motivates us to examine the observed structure and evolution characteristics of MJO before and during its eastward journey across the MC.

Earlier theories attributed the eastward propagation of the MJO to Kelvin-wave dynamics (Lau and Peng 1987; Wang 1988; Wang and Rui 1990). The premise behind the theories was that the free-wave dynamics in the equatorial region (Matsuno 1966) were modulated by diabatic heating (Gill 1980). With the satellite data that became available in the 1970s and the operational reanalysis data in the 1990s, ample evidence supported the existence of convectively coupled moist waves and a distinctive MJO mode with a Kelvin–Rossby wave couplet structure (Wang and Li 1994; Li and Wang 1994; Wheeler and Kiladis 1999).

More recently, the concept of “moisture mode” was put forward. The essence of the moisture mode is the zonal asymmetry of the moisture anomaly to the east of the MJO convection, which serves as a trigger for further development of large-scale convection to the east of the existing MJO convection (Sobel et al. 2001; Fuchs and Raymond 2002, 2005, 2007; Hsu and Li 2012; Sobel and Maloney 2013; Kim et al. 2014). Through observational diagnosis, Hsu and Li (2012) demonstrated that the low-level moisture is primarily caused by anomalous vertical advection, which is associated with the zonal asymmetry of low-level convergence induced by both free-atmospheric and air–sea interaction processes.

Kim et al. (2014) investigated the characteristic difference between a group of MJOs that crossed the MC and another group of MJOs that did not and found that the key difference lay in whether or not a suppressed phase of MJO appears to the east of the wet MJO event. In other words, MJO prefers eastward propagation across the MC when the suppressed convective phase appears to the east of the MJO convection. They argued that the negative heating anomaly associated with the eastern strong suppressed phase of outgoing longwave radiation (OLR) would induce a low-level anticyclonic Rossby gyre, which would advect high mean moisture poleward and lead to the increase of moisture and thus moist static energy to the east of the MJO convection. This, in turn, promotes the eastward propagation of the MJO across the MC.

Motivated by the above work, we would like to address the following specific science questions: Is the eastern strong suppressed phase of OLR really a necessary condition for the MJO to cross the MC? If not, what is the fundamental mechanism responsible for the propagating and nonpropagating cases? The remaining part of the paper is organized as follows. The data and methodology are introduced in section 2. The observed structure and evolution characteristics of the propagating and nonpropagating MJO groups and associated moisture budget analyses are presented in section 3. In section 4, we further separate all propagating events into two groups, with and without the strong suppressed phase of OLR ahead, and examine fundamental dynamic differences between the two groups. Finally, conclusions and a discussion are given in section 5.

2. Data and methods

Daily mean satellite-observed OLR from the National Oceanic and Atmospheric Administration (NOAA) during 1979–2008 was employed as a proxy for deep convection in the tropics. In addition, daily atmospheric variables that include specific humidity q, horizontal wind velocity V (zonal u and meridional υ components), and vertical wind velocity ω from the European Centre for Medium-Range Weather Forecast (ECMWF) interim reanalysis (ERA-Interim; ECMWF 2015) during 1979–2008 were used to examine the MJO 3D structure and calculate the moisture budget. All the ERA-Interim variables are at 12 standard pressure levels from 1000 to 250 hPa and are interpolated with a bilinear interpolation algorithm into a 2.5° latitude × 2.5° longitude horizontal grid so that the reanalysis data grid is consistent with the OLR data grid.

A Lanczos bandpass filter was used to obtain the low-frequency background-state (with a 90-day low-pass filter) MJO-scale (20–90-day bandpass filter) component and the higher-frequency (with a 20-day high-pass filter) component.

A total moisture budget equation may be written as follows (Yanai et al. 1973):

 
formula

In the moisture equation above, t is time, p is the atmospheric pressure, Q2 is the measure of apparent moisture sink, and Lυ is the constant of latent heat of vaporization. The moisture tendency is made up of three terms as shown on the right-hand side of Eq. (1): horizontal advection, vertical advection, and moisture sink associated with evaporation and condensation of water vapor. Here, Q2 was calculated based on 6-hourly reanalysis data following the approach proposed by Yanai et al. (1973).

To investigate the intraseasonal moisture tendency, a 20–90-day bandpass filter operator may be applied to Eq. (1) (Hsu and Li 2012) as follows:

 
formula

To examine specific advective processes responsible for the intraseasonal moisture change, the dependent variables, such as q, u, υ, and ω were decomposed into three components—the low-frequency background mean state, the MJO component, and the higher-frequency component:

 
formula

where an overbar denotes the variable with time scale longer than 90 days, a prime the intraseasonal (20–90-day) variable, and an asterisk the higher-frequency (less than 20 days) variable. The three components of all the variables can be substituted into Eq. (2) so that each advection term can be separated into nine terms. By comparing each of the nine terms, one can reveal major processes that cause the intraseasonal moisture change.

3. Structure and evolution characteristics of propagating and nonpropagating MJOs

At first, the propagating and nonpropagating MJO events were defined. A variance analysis of the 20–90-day-filtered OLR during the winter half year [November–April (NDJFMA)] of 1979–2008 showed a maximum center in the Indian Ocean located at 10°S–5°N, 75°–100°E; this box was selected to construct a reference time series of the filtered OLR, and its standard deviation is about 10 W m−2. Only those cases that have anomalous OLR amplitude less than negative one standard deviation were selected. The time of peak minimum OLR was defined as day zero for each selected MJO case. A total of 96 cases were selected according to the method above. Next, we examined the Hovmöller diagrams of the 20–90-day-filtered OLR field from day −20 to day 25 averaged over 10°S–10°N. The continuous eastward-propagating (EP) case was defined as a case when the contour of −10 W m−2 passed over 120°E without any interruption or gap between days. A total of 46 EP cases were selected according to the criterion above. The nonpropagating (NP) case was defined as a case when the contour of −10 W m−2 clearly showed continuous eastward propagation within the equatorial IO (with zonal extent greater than 30° in longitude) but stopped before approaching 120°E. A total of 14 NP cases were picked out. The unselected cases were either short lived or were stationary without clear continuous eastward propagation.

As expected as defined, both the Hovmöller diagrams and the horizontal evolution maps of the composite OLR anomaly for the EP and NP groups show clear distinctions between the two groups (Figs. 1 and 2). For the EP group, there is continuous eastward propagation in the anomalous OLR field. For the NP group, such propagation is clearly seen in the IO but stops prior to approaching 120°E. A great difference arises from day 5 to day 15, during which the large-scale OLR anomaly associated with EP groups continues moving eastward as it crosses the MC, while the OLR anomaly associated with NP group dissipates and stops in the MC (Fig. 2).

Fig. 1.

Hovmöller diagrams of 20–90-day-filtered OLR anomaly averaged over 10°S–10°N from day −20 to day 25 for the (a) EP and (b) NP composite (W m−2). The red solid line denotes 120°E.

Fig. 1.

Hovmöller diagrams of 20–90-day-filtered OLR anomaly averaged over 10°S–10°N from day −20 to day 25 for the (a) EP and (b) NP composite (W m−2). The red solid line denotes 120°E.

Fig. 2.

Evolutions of 2D 20–90-day-filtered OLR (W m−2) and 850-hPa wind fields (m s−1) from day −20 to day 25 day for the (a) EP and (b) NP composite. Day 0 denotes the time when maximum MJO convection appears in the eastern equatorial IO (yellow box). The yellow solid line denotes 120°E.

Fig. 2.

Evolutions of 2D 20–90-day-filtered OLR (W m−2) and 850-hPa wind fields (m s−1) from day −20 to day 25 day for the (a) EP and (b) NP composite. Day 0 denotes the time when maximum MJO convection appears in the eastern equatorial IO (yellow box). The yellow solid line denotes 120°E.

Why is the anomalous OLR amplitude maintained during its crossing of the MC in one case but it rapidly decays in the other case? To investigate the cause of this difference, we first examined the moisture vertical profiles. According to previous studies (e.g., Sperber 2003; Hsu and Li 2012), the specific humidity anomaly associated with MJO has a tilted vertical structure, with positive anomalies to the east of the MJO convection. What happens to the tilted moisture structure for the EP and NP groups? Figure 3 shows the evolutions of vertical moisture structures from day −10 to day 20 at a 5-day interval. As expected, in both the groups over the IO, the moisture tilting is clearly presented. However, the moisture evolution becomes distinctively different at days 5–15 when the perturbations pass the MC. Whereas the moisture profile in EP groups remains tilted as it smoothly crosses the MC, the tilting structure diminishes in NP groups as a negative humidity anomaly moves westward from 180° at day 5 to 160°E at day 15. We speculate that the westward-moving dry air may hinder the development of new convection to the east of the MJO convective center, preventing its eastward propagation.

Fig. 3.

The vertical–longitude cross sections of 20–90-day-filtered specific humidity (105 kg kg−1) from day −10 to day 20 for (a) EP and (b) NP composite. The black box denotes the position of convective center on the day 0. The red solid line denotes 120°E. The green arrow marks the westward moving center of the anomaly of dry air from day 5 to day 15.

Fig. 3.

The vertical–longitude cross sections of 20–90-day-filtered specific humidity (105 kg kg−1) from day −10 to day 20 for (a) EP and (b) NP composite. The black box denotes the position of convective center on the day 0. The red solid line denotes 120°E. The green arrow marks the westward moving center of the anomaly of dry air from day 5 to day 15.

To examine whether or not the westward-propagating dry signals are real, the specific humidity anomalies averaged from 850 to 700 hPa (according to vertical profiles in Fig. 3) were plotted to show the horizontal structure and evolution from day 0 to day 15 for each NP case. It was found that, for 11 out of the 14 NP cases, there are clear westward-propagating dry signals. Figure 4 shows the composite humidity and 850-hPa wind evolution maps from day 0 to day 15 from the 11 cases. It appears that the dry signal has a Rossby wave–like structure, and the negative specific humidity anomaly center is approximately in phase with low-level anticyclonic gyres.

Fig. 4.

The horizontal patterns of 20–90-day-filtered specific humidity (105 kg kg−1) averaged from 700 to 850 hPa and 20–90-day-filtered 850-hPa wind (m s−1) from day 0 to day 15 derived from 11 NP case composite. The yellow letter A denotes anomalous anticyclonic circulation.

Fig. 4.

The horizontal patterns of 20–90-day-filtered specific humidity (105 kg kg−1) averaged from 700 to 850 hPa and 20–90-day-filtered 850-hPa wind (m s−1) from day 0 to day 15 derived from 11 NP case composite. The yellow letter A denotes anomalous anticyclonic circulation.

To explore the possible source of the westward-propagating dry signals, we extracted the Rossby-wave (RW) signals from raw moisture and wind databased on a wavenumber–frequency analysis, following Wheeler and Kiladis (1999). Westward-propagating signals with a period of 10–30 days and zonal wavenumbers 2–10 were obtained.

Figure 5 shows the composite evolution maps of the RW signals from day 0 to day 15. There are clear westward-propagating RW signals with a pair of low-level anticyclones. Dry anomalies are located to the east of the anticyclone, as anomalous flows advect dry air from higher latitudes equatorward. The anomalous anticyclonic centers appear over 20°S and 25°N at day 0, propagating westward from 200°E at day 0 to 140°E at day 15. The northern branch of the RW appears greater than the southern one (Fig. 5). It is noticed that the dry anticyclonic signals appear to be strengthening from day 0 to day 10. At day 15, the pair of dry anticyclones appears weakening. The wavenumber–frequency analysis confirms that the westward-propagating dry signals originated from equatorial Rossby waves. The westward-moving RW in the equatorial Pacific may act as an independent signal that prevents the eastward propagation of MJO convection across the MC.

Fig. 5.

The horizontal patterns of the Rossby wave of specific humidity (105 kg kg−1) averaged from 700 to 850 hPa and the Rossby wave of 850-hPa wind (m s−1) from day 0 to day 15 derived from the 11 NP case composite. The yellow letter A denotes anomalous anticyclonic circulation.

Fig. 5.

The horizontal patterns of the Rossby wave of specific humidity (105 kg kg−1) averaged from 700 to 850 hPa and the Rossby wave of 850-hPa wind (m s−1) from day 0 to day 15 derived from the 11 NP case composite. The yellow letter A denotes anomalous anticyclonic circulation.

Therefore, the observational analysis indicates the important role of the westward-propagating Rossby wave–type dry precursor signals in the central and western Pacific in prohibiting the eastward propagation of the MJO across the MC. This dry air intrusion effect was clearly seen from the column-integrated moisture tendency profiles averaged during days 5–10 (Fig. 6). During that period, the large-scale convection (i.e., anomaly OLR) centers are located near 120°E. The low-level wind field shows a typical MJO circulation structure, with two Rossby wave cyclonic gyres appearing west of the convection and anomalous easterlies associated with the Kelvin-wave response appearing east of the convection (Figs. 6a,c). Both the OLR and wind anomalies appear weaker in NP than in EP cases. The vertically integrated specific humidity and moist static energy (MSE) tendency fields show a distinctive feature between EP and NP cases. In EP cases, a positive tendency of the moisture and MSE occurs to the east of the convective center (130°–170°E; Fig. 6b). In contrast, a negative tendency appears in NP cases over the same longitudinal band (Fig. 6d). Because of the distinctive difference in the moisture and MSE tendency, the former favors the continuous eastward propagation of the MJO convective system, while the latter prohibits the development of new convection to the east of the existing convective center of the MJO.

Fig. 6.

(top) Horizontal patterns of 20–90-day-filtered OLR (W m−2) and 850-hPa wind (m s−1) fields for the (a) EP and (c) NP composite. (bottom) The zonal distribution of MSE tendency (s−3 kg; red dash line), q tendency (multiplied by 2 × 106 s−1 kg m−2; red solid line), MSE (10−6 s−2 kg; black dash line), q (multiplied by 2 kg m−2; black solid line), and OLR (multiplied by 0.8 W m−2; blue solid line) averaged over 10°S–10°N at days 5–10 for the (b) EP and (d) NP composite.

Fig. 6.

(top) Horizontal patterns of 20–90-day-filtered OLR (W m−2) and 850-hPa wind (m s−1) fields for the (a) EP and (c) NP composite. (bottom) The zonal distribution of MSE tendency (s−3 kg; red dash line), q tendency (multiplied by 2 × 106 s−1 kg m−2; red solid line), MSE (10−6 s−2 kg; black dash line), q (multiplied by 2 kg m−2; black solid line), and OLR (multiplied by 0.8 W m−2; blue solid line) averaged over 10°S–10°N at days 5–10 for the (b) EP and (d) NP composite.

It is noted that the MSE tendency is primarily controlled by the moisture tendency, which can be seen clearly in Figs. 6b and 6d. This prompts us to focus on the humidity tendency. To further investigate specific physical processes responsible for the distinctive moisture tendency difference, we conducted a moisture budget analysis for both the EP and NP groups.

Because −(Q2/Lυ) is calculated as the residual of the moisture in Eq. (1), the budget is exactly balanced in this paper. Figure 7a shows the column-integrated moisture budget from 1000 to 250 hPa averaged during days 5–10 over the east of the convective center. Note that for EP events, the positive q tendency is primarily determined by anomalous vertical advection, as in Hsu and Li (2012), whereas for NP events, anomalous horizontal advection is the major process responsible for the negative tendency. While the anomalous vertical advection in the EP events is much larger than in the NP events, there is an almost equally large negative tendency from the moisture sink in the EP events, which is much weaker in the NP events. Thus, the small difference between vertical advection and moisture sink determines the overall tendency in the EP events. Although horizontal advection in the NP events is the largest negative tendency, positive tendency due to anomalous vertical advection is even larger and is largely offset by the negative moisture sink term.

Fig. 7.

Column-integrated (1000–250 hPa) specific humidity tendency averaged in the region of 10°–10°N, 130°–170°E at days 5–10: (a) terms 1–5 represent, respectively, q tendency, anomalous horizontal advection, anomalous vertical advection, moisture sink of evaporation and condensation, and sum of terms 2–4 (105 s−1 kg m−2) for the EP (red bar) and NP (blue bar) composite; (b) terms 1–9 represent, respectively, anomalous vertical advection terms , , , , , , , , and for the EP (red bars) and NP (blue bars) composite. Column-integrated (1000–700 hPa) moisture divergence averaged in the region of 10°S–10°N, 130°–170°E at days 5–10: (c) terms 1–9 represent, respectively, anomalous moisture divergence terms , , , , , , , , and .

Fig. 7.

Column-integrated (1000–250 hPa) specific humidity tendency averaged in the region of 10°–10°N, 130°–170°E at days 5–10: (a) terms 1–5 represent, respectively, q tendency, anomalous horizontal advection, anomalous vertical advection, moisture sink of evaporation and condensation, and sum of terms 2–4 (105 s−1 kg m−2) for the EP (red bar) and NP (blue bar) composite; (b) terms 1–9 represent, respectively, anomalous vertical advection terms , , , , , , , , and for the EP (red bars) and NP (blue bars) composite. Column-integrated (1000–700 hPa) moisture divergence averaged in the region of 10°S–10°N, 130°–170°E at days 5–10: (c) terms 1–9 represent, respectively, anomalous moisture divergence terms , , , , , , , , and .

A further separation of the horizontal and vertical advection terms indicates that the dominant vertical advection in EP events is primarily attributed to the advection of mean moisture by anomalous vertical velocity (Fig. 7b), which is ultimately linked to the planetary boundary layer convergence (Hsu and Li 2012). The vertically integrated (1000–700 hPa) moisture convergence appears in both the EP and NP events as shown in Fig. 7c. The former is about 4 times as large in EP events as in NP events, which is consistent with the fact that the vertical moisture advection in EP cases is much larger than in NP cases (Fig. 7b) due to relatively stronger intraseasonal ascending motion anomaly to the east of the MJO convection.

The budget analysis above also indicates that it is the dry horizontal advection that plays a crucial role in dissipating and stopping the propagation of the MJO convection in the NP cases. A further examination of the anomalous horizontal advection in NP cases shows that zonal advection dominates meridional advection (Fig. 8a). The diagnosis of the anomalous zonal advection shows that dominated terms are 1) advection of the intraseasonal moisture anomaly by the mean easterly and 2) advection of the mean moisture by the intraseasonal wind (Fig. 8b). To show clearly the cause of anomalous advection, the relationship between the anomalous wind and the mean moisture or the anomalous moisture and the mean wind at the level of maximum moisture advection was examined. Figure 9a shows how the intraseasonal moisture anomaly is advected by the mean wind field. A large-scale negative specific humidity anomaly appears near the date line during days 5–10. This dry anomaly is associated with the westward-propagating Rossby wave–type dry signals identified in Figs. 3 and 4. Figure 9b shows how the intraseasonal wind anomaly advects the mean moisture. Because the mean moisture field is lower toward the eastern Pacific cold tongue, the anomalous low-level easterly to the east of the MJO convection advects dry mean air westward, causing the decrease of the anomalous MSE in the red box region.

Fig. 8.

Column-integrated (1000–250 hPa) specific humidity tendency averaged in the region of 10°S–10°N, 130°–170°E, at days 5–10: (a) terms 1–3 represent, respectively, q horizontal advection, u component, υ component of anomalous q horizontal advection (105 s−1 kg m−2) for the NP composite; (b) terms 1–9 represent, respectively, the anomalous u component of horizontal advection terms , , , , , , , , and for the NP composite.

Fig. 8.

Column-integrated (1000–250 hPa) specific humidity tendency averaged in the region of 10°S–10°N, 130°–170°E, at days 5–10: (a) terms 1–3 represent, respectively, q horizontal advection, u component, υ component of anomalous q horizontal advection (105 s−1 kg m−2) for the NP composite; (b) terms 1–9 represent, respectively, the anomalous u component of horizontal advection terms , , , , , , , , and for the NP composite.

Fig. 9.

(a) The background mean wind (m s−1) and 20–90-day-filtered specific humidity (104 kg kg−1) at 600 hPa; (b) 20–90-day-filtered wind (m s−1) and the background mean specific humidity (104 kg kg−1) at 750 hPa averaged at days 5–10 for the NP composite.

Fig. 9.

(a) The background mean wind (m s−1) and 20–90-day-filtered specific humidity (104 kg kg−1) at 600 hPa; (b) 20–90-day-filtered wind (m s−1) and the background mean specific humidity (104 kg kg−1) at 750 hPa averaged at days 5–10 for the NP composite.

To examine the statistical significance between EP and NP cases, three indices—the 20–90-day-filtered OLR anomaly, the tendency of the vertically integrated (1000–250 hPa) specific humidity anomaly, and the vertically integrated (1000–250 hPa) specific humidity anomaly averaged over 10°S–10°N, 130°–170°E during days 5–10—have been examined with the Monte Carlo simulating method. Two averages can be calculated by separately drawing two random samples from all 60 cases of the set composed of 46 EP and 14 NP cases, and the sizes of two drawn samples match separately the numbers from the 46 EP cases and 14 NP cases. The difference between the two averages from the randomly drawn samples is calculated. This process was repeated 105 times, and then all the differences were ranked in order from smallest to largest. Finally, the difference between the average of all EP cases and the average of all NP cases is statistically significant if the difference is larger than the 95 000th difference or smaller than the 5000th difference. It is shown that all the three indices are statistically significant at the 5% level for the EP and NP groups (Table 1).

Table 1.

Monte Carlo test for the 20–90-day anomaly of OLR (OLR′), q tendency (qtend), and q′ index averaged during days 5–10 in 10°S–10°N, 130°–170°E for the EP–NP group. Shown is the 95% confidence interval of difference between two samples for 105 times roundly sampling. The boldface values are statistically significant.

Monte Carlo test for the 20–90-day anomaly of OLR (OLR′), q tendency (q′tend), and q′ index averaged during days 5–10 in 10°S–10°N, 130°–170°E for the EP–NP group. Shown is the 95% confidence interval of difference between two samples for 105 times roundly sampling. The boldface values are statistically significant.
Monte Carlo test for the 20–90-day anomaly of OLR (OLR′), q tendency (q′tend), and q′ index averaged during days 5–10 in 10°S–10°N, 130°–170°E for the EP–NP group. Shown is the 95% confidence interval of difference between two samples for 105 times roundly sampling. The boldface values are statistically significant.

4. Is the suppressed phase to the east of the MJO a necessary condition for eastward propagation across the MC?

Kim et al. (2014) emphasized the role of the suppressed convective phase of the OLR anomaly to the east of MJO in causing the eastward propagation across the MC. In this section, by separating the EP cases into two groups, one associated with the clear suppressed convective phase of the OLR anomaly and the other without, we intend to examine the fundamental mechanism operating in the eastward propagation scenario.

Among 46 EP cases, a suppressed convection index was introduced to describe the strength of the OLR anomaly to the east of the MJO convection. It was defined as the mean value of 20–90-day-filtered OLR in the region of 10°S–10°N, 120°E–180° at day 0. During that time, the center of the MJO convection is located near 90°E. The 46 EP cases were separated into two subgroups based on the value of their dry indices. An eastward-propagating case with strongly suppressed convective phase (EP-SS) was defined for a case when the suppressed index is greater than the mean value adding to a half of standard deviation (i.e., greater than 9.7 W m−2, above the upper red line in Fig. 10). An eastward-propagating case with weakly suppressed convective phase (EP-WS) was defined for a case when the suppressed index is less than the mean value adding to negative a half of standard deviation (i.e., less than 1.8 W m−2, below the lower red line in Fig. 10). The active convective index in Fig. 10 is based on the mean value of 20–90-day-filtered OLR averaged over the reference box at day 0 shown in Fig. 2. There was a total of 13 EP-SS cases and 12 EP-WS cases selected. There is no systematic relationship between the convective index and the suppressed index in the plot, implying that the positive OLR anomaly does not always appear to the east of the MJO convective center, which means both the EP-SS cases and EP-SW cases, with or without the strong suppressed convective phase of the OLR anomaly to the east of convection, can pass over the MC.

Fig. 10.

The scatterplot of the MJO convection index and the dry index for all 46 EP cases. The upper (lower) red line denotes the suppressed OLR index greater (less) than ½ (−½) standard deviation (W m−2), defined as the EP-SS (EP-WS) suppressed cases.

Fig. 10.

The scatterplot of the MJO convection index and the dry index for all 46 EP cases. The upper (lower) red line denotes the suppressed OLR index greater (less) than ½ (−½) standard deviation (W m−2), defined as the EP-SS (EP-WS) suppressed cases.

As expected from the definition, at day 0 when MJO convection appears at 90°E, there is a clear strongly suppressed convective phase (i.e., positive OLR anomaly) to the east of the MJO convection in the EP-SS composite but a weakly suppressed phase in the EP-WS composite in both the Hovmöller diagram and the evolution pattern of composite OLR anomalies (Figs. 11 and 12). Despite a large difference in the OLR field, the vertically integrated specific humidity tendency fields are quite similar between EP-WS and EP-SS composites (Figs. 13c,d). In both the composites, a positive humidity tendency appears to the east of the convection (120°–170°E) (Figs. 13a–d). This positive tendency ensures the continuous development and eastward propagation of the MJO convection, regardless of the sign of the column-integrated specific humidity anomaly to the east of the convection (Figs. 13c,d).

Fig. 11.

Hovmöller diagrams of 20–90-day-filtered OLR anomaly averaged over 10°S–10°N from day −20 to day 25 for the (a) EP-WS and (b) EP-SS composite (W m−2).

Fig. 11.

Hovmöller diagrams of 20–90-day-filtered OLR anomaly averaged over 10°S–10°N from day −20 to day 25 for the (a) EP-WS and (b) EP-SS composite (W m−2).

Fig. 12.

Evolution of 2D 20–90-day-filtered OLR (W m−2) and 850-hPa wind (m s−1) from day −10 to day 10 for the (a) EP-WS and (b) EP-SS composite.

Fig. 12.

Evolution of 2D 20–90-day-filtered OLR (W m−2) and 850-hPa wind (m s−1) from day −10 to day 10 for the (a) EP-WS and (b) EP-SS composite.

Fig. 13.

(top) The vertical–longitude cross sections of 20–90-day-filtered q tendency (multiplied by ⅓ × 1010 kg kg−1 s−1) averaged over 10°S–10°N at day 0 for the (a) EP-WS and (b) EP-SS composite. (middle) Longitudinal distribution of 20–90-day-filtered OLR (W m−2; blue line), column-integrated (1000–250 hPa) intraseasonal q tendency (108 s−1 kg m−2; red line) and column-integrated specific humidity anomaly (102 kg m−2; black line) averaged over 10°S–10°N at day 0 for the (c) EP-WS and (d) EP-SS composite. (bottom) The vertical–longitude sections of 20–90-day-filtered pressure vertical velocity (ω) averaged over 10°S–10°N (102 kg m s−3) at day 0 for the (e) EP-WS and (f) EP-SS composite.

Fig. 13.

(top) The vertical–longitude cross sections of 20–90-day-filtered q tendency (multiplied by ⅓ × 1010 kg kg−1 s−1) averaged over 10°S–10°N at day 0 for the (a) EP-WS and (b) EP-SS composite. (middle) Longitudinal distribution of 20–90-day-filtered OLR (W m−2; blue line), column-integrated (1000–250 hPa) intraseasonal q tendency (108 s−1 kg m−2; red line) and column-integrated specific humidity anomaly (102 kg m−2; black line) averaged over 10°S–10°N at day 0 for the (c) EP-WS and (d) EP-SS composite. (bottom) The vertical–longitude sections of 20–90-day-filtered pressure vertical velocity (ω) averaged over 10°S–10°N (102 kg m s−3) at day 0 for the (e) EP-WS and (f) EP-SS composite.

The cause of the negative specific humidity to the east of the convection in EP-SS cases is attributed to the dry advection of the anomalous descending motion associated with a positive OLR anomaly in situ (Fig. 13f). This feature is in great contrast to EP-WS cases, in which there is anomalous ascending motion to the east of the MJO convection, and the ascending anomaly advected the mean moisture upward and led to a positive column-integrated moisture anomaly (Fig. 13e).

A column-integrated moisture budget analysis was carried out to reveal why there are positive moisture tendencies in both EP-SS and EP-WS cases, even though anomalous vertical motion has an opposite sign. The result shows that different processes dominate the two scenarios. In EP-SS cases, anomalous horizontal advection and apparent moisture source are dominant terms (Fig. 14a). The former is mainly contributed by the meridional advection, consistent with Kim et al. (2014). The latter tends to offset with anomalous vertical advection. In EP-WS cases, the moistening is primarily caused by anomalous vertical advection, while the horizontal advection also plays a role (Fig. 14a). A further separation of the anomalous vertical advection term shows that the dominant contribution comes from the advection of the mean moisture by anomalous vertical velocity (Fig. 14b).

Fig. 14.

As in Figs. 8, but at day 0 averaged over 10°S–10°N, 120°–170°E. The blue bars are for the EP-WS, and the red bars are for the EP-SS composite. (105 kg m−2).

Fig. 14.

As in Figs. 8, but at day 0 averaged over 10°S–10°N, 120°–170°E. The blue bars are for the EP-WS, and the red bars are for the EP-SS composite. (105 kg m−2).

It is interesting to note that the anomalous meridional moisture advection in EP-SS cases is 5 times as large as that in EP-WS cases, while anomalous zonal advection has the same magnitude (Fig. 15a). The difference in the meridional advection between EP-SS and EP-WS cases is not attributed to the advection of the mean moisture by the intraseasonal meridional wind but is attributed to the advection of high-frequency moisture by high-frequency wind (Fig. 15b). In other words, it is the effect of nonlinear eddy moisture transport that made a difference.

Fig. 15.

Column-integrated (1000–250 hPa) specific humidity tendency averaged in the region of 10°S–10°N, 120°–170°E at day 0: (a) terms 1–3 represent, respectively, q horizontal advection, u component, and υ component of anomalous q horizontal advection (105 s−1 kg m−2) for the EP-SW (blue bars) and EP-SS (red bars) composite; (b) terms 1–9 represent, respectively, the anomalous υ component of horizontal vertical advection terms , , , , , , , , and for EP composite.

Fig. 15.

Column-integrated (1000–250 hPa) specific humidity tendency averaged in the region of 10°S–10°N, 120°–170°E at day 0: (a) terms 1–3 represent, respectively, q horizontal advection, u component, and υ component of anomalous q horizontal advection (105 s−1 kg m−2) for the EP-SW (blue bars) and EP-SS (red bars) composite; (b) terms 1–9 represent, respectively, the anomalous υ component of horizontal vertical advection terms , , , , , , , , and for EP composite.

The advection of the mean background moisture by the intraseasonal meridional wind anomaly was emphasized by Kim et al. (2014). However, as shown in Fig. 15, regardless of EP-SS or EP-WS scenarios, this advective effect is always present. This implies that this advective effect does not depend on the strength of the suppressed phase to the east of the MJO convection. Figure 16 shows the horizontal patterns of the intraseasonal wind anomaly and the mean moisture fields at day 0 in both cases. The Rossby wave–gyre-type circulation patterns with marked poleward flows are seen in both the EP-SS and EP-WS cases. The major difference between them lies in the longitudinal location—the poleward flows shift farther to the west in EP-SS cases. These anomalous flows tend to transport the high mean moisture (which is located near the equator) toward the extratropical region, leading to the buildup of positive moisture and MSE anomalies to the east of the MJO convection.

Fig. 16.

Horizontal patterns of 20–90-day-filtered wind (m s−1) and mean specific humidity (; 102 kg kg−1) fields at 800 hPa on day 0 for the (a) EP-WS and (b) EP-SS composite.

Fig. 16.

Horizontal patterns of 20–90-day-filtered wind (m s−1) and mean specific humidity (; 102 kg kg−1) fields at 800 hPa on day 0 for the (a) EP-WS and (b) EP-SS composite.

It is found that the maximum contribution of eddy moisture transport averaged in all EP-SS cases appears at 800 hPa. To illustrate how high-frequency eddies contribute to the intraseasonal moisture advection, we calculated the dominant EOF pattern of the eddy specific humidity meridional gradient at 800 hPa and associated high-frequency wind fields from day −5 to day 5 for all EP-SS cases. Figure 17 shows the dominant EOF patterns of the high-frequency moisture gradient and associated low-level wind fields. A negative correlation between the north–south moisture gradient and the meridional wind component implies a positive eddy moisture transport and thus an increase of the intraseasonal moisture field during the period. Our result indicates that such a negative correlation exists only in EP-SS, but not in EP-WS cases (figure not shown). A further study is needed to understand how the high-frequency moisture and wind fields form and what controls their structures. It should be noted that the high-frequency term in EP-SS cases exhibits a positive contribution, which is opposite to Hsu and Li (2012). The difference is attributed to different bands used [Hsu and Li (2012) used a 3–10-day high-pass filter, while the present study uses a 20-day bandpass filter] or different sample data used [the present study separates the EP-WS and EP-SS groups, while Hsu and Li (2012) did not]. An in-depth analysis is needed to reveal this difference.

Fig. 17.

The dominant EOF pattern of eddy-specific humidity meridional gradient ∂q*/∂y (shading; 1010 kg kg−1 m−1) and regressed eddy wind field (V*; m s−1) at 800 hPa. The EOF analysis was done for the high-frequency ∂q*/∂y fields from day −5 to day 5 for all EP-SS cases. The eddy wind field is regressed to the principle component of the dominant ∂q*/∂y EOF mode.

Fig. 17.

The dominant EOF pattern of eddy-specific humidity meridional gradient ∂q*/∂y (shading; 1010 kg kg−1 m−1) and regressed eddy wind field (V*; m s−1) at 800 hPa. The EOF analysis was done for the high-frequency ∂q*/∂y fields from day −5 to day 5 for all EP-SS cases. The eddy wind field is regressed to the principle component of the dominant ∂q*/∂y EOF mode.

The same statistical significance test as in the EP and NP groups was applied to the EP-WS and EP-SS groups for the same three indices averaged in 10°S–10°N, 120°–170°E on the zeroth day, as shown in Table 2. For the EP-WS and EP-SS groups, the OLR and specific humidity anomalies are statistically significant at the 95% confidence level, but the moisture tendency is not. This result is consistent with the fact that a positive humidity tendency appears in both groups.

Table 2.

Monte Carlo test for anomaly of OLR′, qtend, and q′ index on the zeroth day averaged in 10°S–10°N, 120°–170°E for the EP-WS–EP-SS group. Shown is the 95% confidence interval of difference between two samples for 105 times roundly sampling. The boldface values are statistically significant.

Monte Carlo test for anomaly of OLR′, q′tend, and q′ index on the zeroth day averaged in 10°S–10°N, 120°–170°E for the EP-WS–EP-SS group. Shown is the 95% confidence interval of difference between two samples for 105 times roundly sampling. The boldface values are statistically significant.
Monte Carlo test for anomaly of OLR′, q′tend, and q′ index on the zeroth day averaged in 10°S–10°N, 120°–170°E for the EP-WS–EP-SS group. Shown is the 95% confidence interval of difference between two samples for 105 times roundly sampling. The boldface values are statistically significant.

5. Conclusions and discussion

Observed OLR and 30-yr ERA-Interim data were used to understand the fundamental processes that control the MJO propagation across the Maritime Continent (MC). A total of 46 eastward-propagating (EP) MJO cases were selected based on the Hovmöller diagrams of the 20–90-day-filtered OLR field averaged over 10°S–10°N, in which the contour of −10 W m−2 extends from the equatorial central IO all the way to 120°E without any interruption or gap. The nonpropagating (NP) cases were selected from cases when the contour of −10 W m−2 shows clearly continuous eastward propagation within the equatorial Indian Ocean (with zonal extent greater than 30° in longitude) but stops before approaching 120°E. A total of 14 NP cases were picked out.

An examination of horizontal and vertical structures of the EP and NP composites shows that a critical difference lies in the column-integrated moisture tendency to the east of the MJO convection. That is, when the MJO convection arrives near 120°E longitude, there is a positive column-integrated moisture tendency to the east in the EP composite but a negative tendency in the NP composite. The cause of the negative tendency in NP events is primarily attributed to a dry air intrusion mechanism through which a westward-propagating Rossby wave–type dry signal dissipates the MJO convection over the Maritime Continent. A column-integrated moisture budget analysis further confirms this dry air dissipating mechanism. The negative specific humidity tendency in NP events is primarily attributed to the advection of anomalous moisture by the mean easterly wind and the advection of dry mean moisture by anomalous easterlies.

It is interesting to note that, while there is a clear difference in the moisture tendency to the east of the MJO convection between the EP events (positive tendency) and the NP events (negative tendency), the sign of the moisture anomaly is still the same (positive) in both EP and NP events. Given that the moisture mode theory (Hsu and Li 2012; Sobel and Maloney 2013) only requires a positive moisture anomaly to the east of the MJO convection, the observational result challenges whether or not the moisture mode dynamics work here. We hypothesize that there must be a critical value (threshold) of the moisture anomaly. Note that the moisture anomaly in the NP events, while still positive, is certainly of lower amplitude than the moisture anomaly in the EP events. It is likely that convection can only occur when the convective instability criterion or, equivalently, the low-level moisture anomaly exceeds a critical value (threshold). For different regions, this moisture threshold may be different as the mean state changes. Another possible mechanism for nonpropagation in NP events is attributed to westward-propagating dry signals in the equatorial Pacific during days 0–15. It is found that the source of the dry signal arises from westward-propagating equatorial Rossby waves that have a twin-anticyclone structure. The dry Rossby-wave signals appear independent of the MJO and may play an important role in hindering the development of new convection to the east of the MJO convection and preventing its eastward propagation.

By further separating eastward-propagating cases into two groups, one with and the other without the clear suppressed phase of OLR to the east of the MJO convection, we attempted to examine fundamental differences between the two groups. It is noted that in the strong suppressed phase group, a negative column-integrated specific humidity anomaly and a descending motion anomaly appear to the east of the MJO convection. This feature is opposite to the weakly suppressed phase group, in which there are a positive column-integrated specific humidity anomaly and an ascending motion anomaly to the east of the MJO convection. Nevertheless, a positive column-integrated moisture tendency appears in both the composites. It is this positive moisture tendency that promotes the continuous eastward propagation in both groups.

The diagnosis of the column-integrated moisture budget in both groups shows that processes controlling the moisture tendency differ markedly. In the weak suppressed composite, the major moistening process is anomalous vertical advection, while in the strong suppressed composite, the major process is anomalous horizontal advection, such as the advection of high-frequency moisture by high-frequency meridional wind and the advection of mean moisture by intraseasonal wind anomalies.

It is interesting to note that anomalous advection of mean moisture by intraseasonal meridional wind in the EP-SS and EP-WS composite has a similar value. This poses an interesting question: namely, what causes poleward low-level meridional wind anomalies in EP-WS cases? It is easily understood that dry air associated with a large positive OLR anomaly in EP-SS cases would induce a negative heating anomaly and thus a low-level anticyclonic Rossby-wave response to the west of the heating anomaly, as pointed out by Kim et al. (2014). In the presence of a nearly normal OLR anomaly to the east of the MJO convection in EP-WS cases, poleward anomalous flows are not expected. We hypothesize that it is likely a result of anticyclonic shear of zonal wind associated with Kelvin-wave response to the MJO heating. With a maximum easterly anomaly on the equator associated with the Kelvin-wave response, the curl of the zonal wind anomaly would induce anticyclonic shear at both sides of the equator. This anticyclonic shear could induce boundary layer divergence, which could decrease local moisture. (Our diagnosis shows that indeed there is anomalous divergence off the equator in the atmospheric boundary layer in EP-WS cases.) The reduced moisture can further suppress midtropospheric heating and induce low-level anticyclonic flows.

The fact that MJO propagates across the Maritime Continent regardless of descending dry air to the east of the convection suggests that the dry air is not a necessary condition for MJO to cross the Maritime Continent. An important precursor condition is the westward-propagating Rossby-wave dry signal in the equatorial Pacific, based on the analysis result in section 3. Thus, a special attention should be paid to the westward-propagating dry Rossby-wave signal in the equatorial central and western Pacific. Both detection and initialization of such low-frequency dry signals appear critical for improving the operational model forecast of the MJO.

Acknowledgments

This research was supported by China National 973 Project 2015CB453200, NSFC Grant 41475084/41375095, ONR Grant N00014-1210450, the Jiangsu Shuang-Chuang Team, and by the IPRC, which is sponsored by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC).

REFERENCES

REFERENCES
Chen
,
T.-C.
, and
M.
Murakami
,
1988
:
The 30–50 day variation of convective activity over the western Pacific Ocean with emphasis on the northwestern region
.
Mon. Wea. Rev.
,
116
,
892
906
, doi:.
Chen
,
T.-C.
,
R. Y.
Tzeng
, and
M. C.
Yen
,
1988
:
Development and life cycle of the Indian monsoon: Effect of the 30–50 day oscillation
.
Mon. Wea. Rev.
,
116
,
2183
2199
, doi:.
Ding
,
Q.-H.
, and
B.
Wang
,
2007
:
Intraseasonal teleconnection between the summer Eurasian wave train and the Indian monsoon
.
J. Climate
,
20
,
3751
3767
, doi:.
ECMWF
,
2015
: ERA Interim, Daily. ECMWF. Subset used: 1979–2008, accessed December 2013. [Available online at http://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=pl.]
Fu
,
X.
,
J.-Y.
Lee
,
P.-C.
Hsu
,
H.
Taniguchi
,
B.
Wang
,
W.
Wang
, and
S.
Weaver
,
2013
:
Multi-model MJO forecasting during DYNAMO/CINDY period
.
Climate Dyn.
,
41
,
1067
1081
, doi:.
Fuchs
,
Z.
, and
D. J.
Raymond
,
2002
:
Large-scale modes of a nonrotating atmosphere with water vapor and cloud–radiation feedbacks
.
J. Atmos. Sci.
,
59
,
1669
1679
, doi:.
Fuchs
,
Z.
, and
D. J.
Raymond
,
2005
:
Large-scale modes in a rotating atmosphere with radiative–convective instability and WISHE
.
J. Atmos. Sci.
,
62
,
4084
4094
, doi:.
Fuchs
,
Z.
, and
D. J.
Raymond
,
2007
:
A simple, vertically resolved model of tropical disturbances with a humidity closure
.
Tellus
,
59A
,
344
354
, doi:.
Gill
,
A. E.
,
1980
:
Some simple solutions for heat-induced tropical circulation
. Quart. J. Roy. Meteor. Soc., 106, 447–462, doi:.
Hartmann
,
D. L.
, and
M. L.
Michelsen
,
1989
:
Intraseasonal periodicities in Indian rainfall
.
J. Atmos. Sci.
,
46
,
2838
2862
, doi:.
Hsu
,
P.-C.
, and
T.
Li
,
2011
:
Interactions between boreal summer intraseasonal oscillations and synoptic-scale disturbances over the western North Pacific. Part II: Apparent heat and moisture sources and eddy momentum transport
.
J. Climate
,
24
,
942
961
, doi:.
Hsu
,
P.-C.
, and
T.
Li
,
2012
:
Role of the boundary layer moisture asymmetry in causing the eastward propagation of the Madden–Julian oscillation
.
J. Climate
,
25
,
4914
4931
, doi:.
Hsu
,
P.-C.
,
T.
Li
, and
H.
Murakami
,
2014
:
Moisture asymmetry and MJO eastward propagation in an aquaplanet general circulation model
.
J. Climate
,
27
,
8747
8760
, doi:.
Kim
,
D.
, and Coauthors
,
2009
:
Application of MJO simulation diagnostics to climate models
.
J. Climate
,
22
,
6413
6436
, doi:.
Kim
,
D.
,
J.
Kug
, and
A.
Sobel
,
2014
:
Propagating versus nonpropagating Madden–Julian oscillation events
.
J. Climate
,
27
,
111
125
, doi:.
Knutson
,
T. R.
, and
K. M.
Weickmann
,
1987
:
30–60 day atmospheric oscillations: Composite life cycles of convection and circulation anomalies
.
Mon. Wea. Rev.
,
115
,
1407
1436
, doi:.
Krishnamurti
,
T. N.
,
1985
:
Summer monsoon experiment—A review
.
Mon. Wea. Rev.
,
113
,
1590
1626
, doi:.
Krishnamurti
,
T. N.
, and
D.
Subrahmanyam
,
1982
:
The 30–50 day mode at 850 mb during MONEX
.
J. Atmos. Sci.
,
39
,
2088
2095
, doi:.
Lau
,
K. M.
, and
P. H.
Chan
,
1986
:
Aspects of the 40–50 day oscillation during the northern summer as inferred from outgoing longwave radiation
.
Mon. Wea. Rev.
,
114
,
1354
1367
, doi:.
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
, doi:.
Li
,
T.
,
2014
: Recent advance in understanding the dynamics of the Madden–Julian oscillation. J. Meteor. Res., 28, 1–33, doi:.
Li
,
T.
, and
B.
Wang
,
1994
:
The influence of sea surface temperature on the tropical intraseasonal oscillation: A numerical study
.
Mon. Wea. Rev.
,
122
,
2349
2362
, doi:.
Li
,
T.
, and
B.
Wang
,
2005
:
A review on the western North Pacific monsoon: Synoptic-to-interannual variabilities
.
Terr. Atmos. Oceanic Sci.
,
16
,
285
314
.
Li
,
T.
, and
C.
Zhou
,
2009
:
Planetary scale selection of the Madden–Julian oscillation
.
J. Atmos. Sci.
,
66
,
2429
2443
, doi:.
Li
,
T.
,
C.
Zhao
,
P.-C.
Hsu
, and
T.
Nasuno
,
2015
:
MJO initiation processes over the tropical Indian Ocean during DYNAMO/CINDY2011
.
J. Climate
,
28
,
2121
2135
, doi:.
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
, doi:.
Madden
,
R. A.
, and
P. R.
Julian
,
1972
:
Description of global-scale circulation cells in the tropics with a 40–50 day period
.
J. Atmos. Sci.
,
29
,
1109
1123
, doi:.
Majda
,
A. J.
, and
S. N.
Stechmann
,
2009
:
A simple dynamical model with features of convective momentum transport
.
J. Atmos. Sci.
,
66
,
373
392
, doi:.
Matsuno
,
T.
,
1966
:
Quasi-geostrophic motions in the equatorial area
.
J. Meteor. Soc. Japan
,
44
,
25
43
.
Matthews
,
A. J.
,
2008
:
Primary and successive events in the Madden–Julian oscillation
.
Quart. J. Roy. Meteor. Soc.
,
134
,
439
453
, doi:.
Murakami
,
T.
, and
T.
Nakazawa
,
1985
:
Tropical 45 day oscillations during the 1979 Northern Hemisphere summer
.
J. Atmos. Sci.
,
42
,
1107
1122
, doi:.
Nakazawa
,
T.
,
1988
:
Tropical super clusters within intraseasonal variations over the western Pacific
.
J. Meteor. Soc. Japan
,
66
,
823
839
.
Sobel
,
A. H.
, and
E. D.
Maloney
,
2013
:
Moisture modes and the eastward propagation of the MJO
.
J. Atmos. Sci.
,
70
,
187
192
, doi:.
Sobel
,
A. H.
,
J.
Nilsson
, and
L. M.
Polvani
,
2001
:
The weak temperature gradient approximation and balanced tropical moisture waves
.
J. Atmos. Sci.
,
58
,
3650
3665
, doi:.
Sperber
,
K. R.
,
2003
:
Propagation and the vertical structure of the Madden–Julian oscillation
.
Mon. Wea. Rev.
,
131
,
3018
3037
, doi:.
Sperber
,
K. R.
,
J. M.
Slingo
,
P. M.
Inness
, and
W. K.-M.
Lau
,
1997
:
On the maintenance and initiation of the intraseasonal oscillation in the NCEP/NCAR reanalysis and in the GLA and UKMO AMIP simulations
.
Climate Dyn.
,
13
,
769
795
, doi:.
Wang
,
B.
,
1988
:
Dynamics of tropical low-frequency waves: An analysis of the moist Kelvin wave
.
J. Atmos. Sci.
,
45
,
2051
2065
, doi:.
Wang
,
B.
, and
H.
Rui
,
1990
:
Dynamics of the coupled moist Kelvin–Rossby wave on an equatorial β-plane
.
J. Atmos. Sci.
,
47
,
397
413
, doi:.
Wang
,
B.
, and
T.
Li
,
1994
:
Convective interaction with boundary-layer dynamics in the development of a tropical intraseasonal system
.
J. Atmos. Sci.
,
51
,
1386
1400
, doi:.
Wang
,
B.
, and
F.
Liu
,
2011
:
A model for scale interaction in the Madden–Julian oscillation
.
J. Atmos. Sci.
,
68
,
2524
2536
, doi:.
Weickmann
,
K. M.
,
1983
:
Intraseasonal circulation and outgoing longwave radiation modes during Northern Hemisphere winter
.
Mon. Wea. Rev.
,
111
,
1838
1858
, doi:.
Weickmann
,
K. M.
,
G. R.
Lussky
, and
J. E.
Kutzbach
,
1985
:
Intraseasonal (30–60 day) fluctuations of outgoing longwave radiation and the 250 mb streamfunction during northern winter
.
Mon. Wea. Rev.
,
113
,
941
961
, doi:.
Wheeler
,
M.
, and
G. N.
Kiladis
,
1999
:
Convectively coupled equatorial waves: Analysis of clouds and temperature in the wavenumber–frequency domain
.
J. Atmos. Sci.
,
56
,
374
399
, doi:.
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
, doi:.
Zhao
,
C.-B.
,
T.
Li
, and
T.
Zhou
,
2013
:
Precursor signals and processes associated with MJO initiation over the tropical Indian Ocean
.
J. Climate
,
26
,
291
307
, doi:.

Footnotes

*

School of Ocean and Earth Science and Technology Contribution Number 9507, International Pacific Research Center Contribution Number 1151, and Earth System Modelling Center Contribution Number 068.

Publisher’s Note: This article was revised on 3 November 2015 to correct errors in the abstract and update the Acknowledgments section.