Abstract

An environment favorable for the development of the Madden–Julian oscillation (MJO) was investigated by classifying MJO-like atmospheric patterns as MJO and regionally confined convective (RCC) events. Comparison of MJO and RCC events showed that even when preceded by a major convective suppression event, convective events did not develop into an MJO when large-scale buildup of moist static energy (MSE) was inhibited. The difference in the MSE accumulation between MJO and RCC is related to the contrasting low-frequency basic-state sea surface temperature (SST) pattern; the MJO and RCC events were associated with anomalously warm and cold low-frequency SSTs prevailing over the western to central Pacific, respectively. Differences in the SST anomaly field were absent from the intraseasonal frequency range of 20–60 days. The basic-state SST pattern associated with the MJO was characterized by a positive zonal SST gradient from the Indian Ocean to the western Pacific, which provided a long-standing condition that allowed for sufficient buildup of MSE across the Indian Ocean to the western Pacific via large-scale low-level convergence over intraseasonal and longer time scales. The results of this study suggest the importance of such a basic-state SST, with a long-lasting positive zonal SST gradient, for enhancing convection over a longer than intraseasonal time scale in realizing a complete MJO life cycle.

1. Introduction

The Madden–Julian oscillation (MJO) is a prominent intraseasonal variability in the tropics, which has been studied extensively since its discovery in 1971 (Madden and Julian 1971). This phenomenon is widely accepted to be an eastward-proceeding envelope of convective activity coupled with circulation (Zhang 2005); however, defining the MJO, and thereby distinguishing it from tentatively organized convection, is still regarded as a difficult problem on its own (Kiladis et al. 2014; Straub 2013). Thus, a number of MJO indices have been proposed depending on the choice of physical variables and methods of index construction (e.g., Wheeler and Hendon 2004, hereafter WH04; Matthews 2008; Wheeler and Weickmann 2001).

Of the existing indices, the Real-time Multivariate MJO index (RMM) (WH04) has been widely applied to identify dates associated with the MJO. The RMM detects large-scale circulation coupled with convection by applying a combined empirical orthogonal function (EOF) method to a combination of de-seasoned outgoing longwave radiation (OLR) and upper-level (200 hPa; U200) and lower-level (850 hPa; U850) zonal wind data. This simple implementation, with applicability for real-time use across all seasons, has made it a standard in MJO analysis. However, the high amplitude of the RMM signifies only that the atmosphere assumes an MJO-like pattern in circulation and convection momentarily. Therefore, to capture the MJO as an entity of eastward-propagating convective activity on an intraseasonal time scale, identification of MJO events using the RMM requires consideration of the continuity of the trajectory of the RMM projection in the RMM phase space. Intermittent occurrences of days with high RMM amplitudes need to be distinguished from the MJO by defining the MJO as a time sequence with an RMM projection trajectory that signifies continuous eastward propagation of convective activity from the Indian Ocean (IO) to the western Pacific (WP).

In contrast to the extensive work that has already been undertaken on the MJO based on the RMM (cf. Zhang 2013 and references therein), there has been limited research regarding what distinguishes the MJO from discontinuous events of RMM amplification. Hirata et al. (2013) identified convective events that are regionally confined, either over the IO or around the Maritime Continent (MC), and suggested that the intensity of convective suppression was an important factor in determining the behavior of subsequent active convection. Kim et al. (2014) also investigated the differences between propagating and nonpropagating MJO, and their results indicated the importance of the development of a strong dry anomaly prior to the propagating MJO. Moreover, precursor signals for MJO initiation that are absent from large-scale convective events with a low RMM amplitude have been documented by Ling et al. (2013), and factors associated with the termination of MJO events at different RMM phases have been investigated by Stachnik et al. (2015).

The MC is also known to be an influential factor in the eastward propagation of MJO convective activities. Convective activities of the MJO often weaken and stagnate over the MC before propagating to the WP (Rui and Wang 1990; Hendon and Salby 1994; Zhang and Ling 2017). Such an effect of the MC is known as the “barrier effect” on MJO propagation. Conditions such as the transition from an overland dominant precipitation regime to an oversea dominant regime (Zhang and Ling 2017), and sufficient horizontal advection of moisture and moist static energy (MSE; Miura et al. 2007; Kim et al. 2014; Feng et al. 2015) have been suggested as important factors for MJO propagation to the WP. In model simulations, other factors such as correct representation of low-level winds (Inness et al. 2003) and the choice of cumulus parameterization (Zhu et al. 2017) have been indicated to be important for successful propagation of the MJO over the MC.

It has been suggested that SST patterns generated by air–sea interaction on an intraseasonal time scale (Shinoda et al. 1998) are also important for a successful MJO (Flatau et al. 1997; Hirata et al. 2013; DeMott et al. 2014; Zhu et al. 2017). However, it is uncertain as to whether there is a background state longer than the intraseasonal time scale that provides long-standing conditions supporting MJO development. This consideration is supported by theoretical work, which shows that a slowly eastward-propagating pattern like the MJO can be reproduced through interaction of wave activities with planetary-scale moisture anomalies (Majda and Stechmann 2009). We hypothesize that such a basic-state condition should exist to allow an MJO-like large-scale atmospheric circulation to develop and to persist on an intraseasonal time scale. In this study, we investigate this hypothesis by focusing on differences in the basic-state SST, which provides a bottom boundary condition for much of the convective region of MJO events, and we reveal essential environmental factors for the MJO development.

2. Data

This study employed the following standard datasets for calculating the RMM: interpolated OLR (Liebmann and Smith 1996) from the National Oceanic and Atmospheric Administration (NOAA), and U850 and U200 from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) Reanalysis-1 (R1; Kalnay et al. 1996). The analysis spanned a period from 1 January 1979 to 31 December 2012 with a horizontal resolution of 2.5° × 2.5°. These choices are made to facilitate comparison with the results of previous studies, and to keep within the range of the final product of the WH04 RMM provided by the Australian Bureau of Meteorology (2018), which ends on 31 December 2013. However, we have also confirmed that extending the analysis period to 31 December 2016 does not change the general conclusion of this study. We also used NCEP–NCAR R1 data for the variables required to calculate the MSE budget terms (cf. section 4b).

The NOAA Optimum Interpolation SST, version 2 (OISST2; Reynolds et al. 2002) was analyzed from 1 January 1982 to 31 December 2012 to investigate the background SST state favorable for MJO development. The resolution of the OISST (0.25° × 0.25°) was reduced to the resolution of the other NCEP–NCAR data (2.5° × 2.5°), by using area averaging. This reduction was validated by analyzing the spatial scale at which SST became coherent with convective activity in the tropics. This was achieved by investigating the resolution at which the SST and the total column water (TCW), which is closely related to convective activity in the tropics (Bretherton et al. 2004), becomes well correlated. For consistency between the TCW and SST within this analysis, we used ECMWF interim reanalysis data (ERA-Interim, hereafter ERA-I) provided in 11 different resolutions between 0.125° × 0.125° and 0.3° × 3.0° for both TCW and SST. The correlation of SST and TCW was calculated between 30°S and 30°N, at resolutions of 0.25° × 0.25°, 1.0° × 1.0°, and 2.5° × 2.5° using the original ERA-I data from 1 January 1982 to 31 December 2015 (Fig. 1). Correlations at lower resolutions at 5.0° × 5.0° and 10.0° × 10.0° were also calculated by area averaging the 2.5° × 2.5° data to those resolutions. Figure 1 shows that the correlation between SST and TCW was low for higher-resolution data. However, the correlation increased above 0.6 when the resolution was reduced below 2.5° × 2.5° resolution. Therefore, we determined that reduction of the resolution of SST data to 2.5° × 2.5° was reasonable for this study, which is intended to examine the effect of SST on convective activities over the tropics.

Fig. 1.

Resolution dependency of the correlation between SST and TCW from the ERA-I over the tropics (30°S–30°N).

Fig. 1.

Resolution dependency of the correlation between SST and TCW from the ERA-I over the tropics (30°S–30°N).

It was noted that the September 1994 OLR data had unrealistically high values with respect to the satellite switching between 16 September and 17 September. Consequently, data from 1 May 1994 to 31 January 1995 were omitted to eliminate data contaminated from satellite switching by the time filtering (cf. section 3).

3. MJO detection method

The purpose of this study is not real-time evaluation of the MJO; hence, we simplified the de-seasoning process of the original WH04 to 20–120-day Lanczos bandpass filtering (Duchon 1979). The filter employed 241 symmetric weights, and 120 days at both ends of the data were truncated. A relatively wide window for the bandpass filter was used to include signals on a near-seasonal time scale in consideration of MJO events that appeared to be driven by the seasonal cycle (Miura et al. 2015). With the exception of utilizing this bandpass- filtered data, the RMM calculation followed that of the original WH04 RMM. The resulting RMM series had a smoother trajectory than the original WH04 RMM series (Fig. 2); however, the overall features were consistent with those of WH04.

Fig. 2.

Examples of MJO (red), IO-RCC (blue), and WP-RCC (green) RMM trajectories. Solid lines indicate RMM trajectories calculated by using the 20–120-day filter, and dotted lines indicate RMM trajectories of the original WH04 RMM trajectories provided by the Australian Bureau of Meteorology. The trajectories shown are during 13 Jan–5 Mar 2002 for the MJO, 7 Jan–17 Jan 1982 for the IO-RCC, and 17 Nov–5 Dec 1989 for the WP-RCC event. The inner circle indicates an RMM amplitude of Ac = 0.8.

Fig. 2.

Examples of MJO (red), IO-RCC (blue), and WP-RCC (green) RMM trajectories. Solid lines indicate RMM trajectories calculated by using the 20–120-day filter, and dotted lines indicate RMM trajectories of the original WH04 RMM trajectories provided by the Australian Bureau of Meteorology. The trajectories shown are during 13 Jan–5 Mar 2002 for the MJO, 7 Jan–17 Jan 1982 for the IO-RCC, and 17 Nov–5 Dec 1989 for the WP-RCC event. The inner circle indicates an RMM amplitude of Ac = 0.8.

We defined the MJO by specifying a set of criteria to be met by the RMM trajectory, the physical implications of which are consistent with the general characteristics of the MJO. The MJO criteria, followed by their concise physical implications in parentheses, are as follows: 1) it proceeds from at least phase 2 to phase 7 (large-scale circulation propagates through the IO to the WP); 2) it does not skip more than one phase(continuous propagation in real space); 3) it does not recede more than one phase(disqualification of major westward-propagating events), 4) the average RMM amplitude exceeds the critical value Ac, and consecutive days with amplitude below Ac are fewer than 15 days (maintenance of typical MJO structure); and 5) the tracking is completed in 20–90 days (completion of an event within an intraseasonal time scale). The value of Ac was subjectively set at 0.8 because a sufficiently strong signal of an MJO-like atmospheric pattern appeared to be present when the RMM amplitude was at least 0.8 (Fig. 3). For example, the RMM amplitude dropped below Ac for 4 days during the MJO event shown in Fig. 3, which corresponds to the MJO trajectory in Fig. 2; however, strong eastward-proceeding convective activity was present even when the RMM amplitude was small. The critical value was set at 0.8 rather than a higher value to maximize the number of MJO events. Moreover, it was confirmed that the general conclusions of this study were insensitive to Ac values between 0.8 and 1.0.

Fig. 3.

The 15°S–15°N averaged U850 (color; m s−1) and OLR (contour; W m−2) time–longitude Hovmöller during an MJO event during 13 Jan–5 Mar 2002. The color bar on the right indicates the daily RMM phase: (red) phase 1, (orange) phase 2, (yellow) phase 3, (light green) phase 4, (blue) phase 5, (purple) phase 6, and (magenta) phase 7. Periods without color indicate days when the RMM amplitude was lower than Ac.

Fig. 3.

The 15°S–15°N averaged U850 (color; m s−1) and OLR (contour; W m−2) time–longitude Hovmöller during an MJO event during 13 Jan–5 Mar 2002. The color bar on the right indicates the daily RMM phase: (red) phase 1, (orange) phase 2, (yellow) phase 3, (light green) phase 4, (blue) phase 5, (purple) phase 6, and (magenta) phase 7. Periods without color indicate days when the RMM amplitude was lower than Ac.

The criteria were relaxed from a strict anticlockwise procession on the RMM phase space to allow for some skipping and receding of RMM phases and for a temporal drop in RMM amplitude. These criteria were implemented in consideration of the existence of small-scale disturbances within the MJO that may cause temporary jumps, reversals, and amplitude drops of the RMM (Fig. 3). Relaxation of the criteria has the effect of making the MJO detection insensitive to unrealistic phase flips. We note that different measures for the same purpose have been implemented in preceding researches (e.g., Matthews 2008).

Following the criteria, MJO events were detected by first identifying those days projected on phase 2 and then tracking backward to phase 1 and forward up to phase 8. Backward tracking terminated when the RMM amplitude fell below Ac or when a phase change occurred. Forward tracking terminated when the RMM amplitude fell below Ac any time after the first day of phase 7. Tracking was also terminated when a phase change occurred after the first day of phase 8. We defined the initiation and termination of the active phase of the MJO as the first and the last day of phase 2 and phase 7, respectively. Because the attribution of an RMM phase is not legitimate when the amplitude is small, phase retreat or procession is considered only when the RMM amplitude exceeds Ac. In this way, 112 MJO events were detected during 1982–2012 with removal of two events from May 1994 to January 1995, the period in which the Lanczos-filtered data were contaminated by data from September 1994.

Days with an RMM amplitude greater than Ac were further categorized into two types of regionally confined convective events (RCC): Indian Ocean (IO)-RCC and western Pacific (WP)-RCC (depending on the location of their occurrence). We defined IO-RCC and WP-RCC as events of RMM amplification above Ac that were not part of MJO events and were restricted to phases 2–4 and phases 5–7, respectively (Fig. 2). Initiation of IO-RCC and WP-RCC was defined by the first day of the RMM amplification within the respective phase range; an event was terminated when the RMM amplitude dropped below Ac at any time after initiation. Any amplification of RMM not included in the above categories (e.g., RMM amplification that spans phases 2–5) was not considered in this research. In this way, we distinguished temporary events of RMM amplification from conventional MJO events, which are usually complemented by amplification of RMM phases in RMM phase ranges of both IO-RCC and WP-RCC. In this study 51 IO-RCC events and 55 WP-RCC events were identified.

The method described in this section is not definitive for identifying MJO events or RCC. However, this method is sufficient for our purpose in designating an adequate number of convective events that demonstrate the conventional properties of MJO as MJO events and those of other regionally confined and short-lived convective events as RCC events. The results obtained, enable their properties to be analyzed and compared statistically. We also note that although RMM is predominantly determined by large-scale circulation (Straub 2013), this method captures large-scale convective activity events in terms of signal strength and spatial scale as adequately as other MJO detecting methods that use only OLR values (e.g., Matthews 2008; Hirata et al. 2013).

4. Results

a. Circulation and convective activities of MJO and RCC

The circulation and convection patterns of MJO and RCC were compared by using composites of anomaly fields for OLR and U850. Anomalies were calculated by subtracting the daily climatology of the period of 1 January 1982–31 December 2011.

Differences in circulation and convection between the MJO and the RCCs are apparent in the composite time–longitude Hovmöller diagrams of the meridionally averaged (15°S–15°N) anomalies of OLR and U850 centered at the initiation of the active phase of the MJO, which is the first day of phase 2; IO-RCC; and WP-RCC (Fig. 4). In the MJO composite, strong signals of both negative OLR and positive U850 anomalies propagated from approximately 60°E to approximately 180°W over 30 days. However, these signals were short lived, weaker, and stagnant for the RCCs. Signals of convective activity with durations of 10–15 days were present only at approximately 60°–100°E for the IO-RCC and approximately 120°E–180° for the WP-RCC. We note that although the applied composite ignored the differences in the propagation speed of MJO events, it captured the eastward propagation of the MJO and clarified the differences between MJO and RCCs. These results fit the objective of this study.

Fig. 4.

Composite Hovmöller of 15°S–15°N averaged anomalies of OLR (contours; W m−2) and U850 (color; m s−1) for the (a) MJO, (b) IO-RCC, and (c) WP-RCC. Negative OLR anomalies are indicated by blue solid lines and positive OLR anomalies are indicated by magenta dotted lines. Contours are at 3 W m−2 intervals.

Fig. 4.

Composite Hovmöller of 15°S–15°N averaged anomalies of OLR (contours; W m−2) and U850 (color; m s−1) for the (a) MJO, (b) IO-RCC, and (c) WP-RCC. Negative OLR anomalies are indicated by blue solid lines and positive OLR anomalies are indicated by magenta dotted lines. Contours are at 3 W m−2 intervals.

In terms of the signal for convective suppression, interesting similarities were noted between the MJO and RCCs (Fig. 4). For example, the convective activities of both the MJO and IO-RCC were preceded by a signal of anomalously positive OLR that slowly propagated eastward from approximately 60°E to 180° over 30 days and ended with displacement or diminishment of convective activity over the IO (Figs. 4a and 4b). Furthermore, the amplitude of the positive OLR anomaly found in the IO-RCC is comparable to that of the MJO over the IO and WP regions. On the contrary, the WP-RCC convective activity over the WP did not appear to be preceded by significant convective suppression over the same region. However, the WP-RCC convective activities were accompanied by a strong signal of positive OLR over the IO region that continued to propagate eastward to the WP well after the convective activity had diminished (Fig. 4c). We also noted that the magnitude of convective suppression in the WP-RCC was significantly stronger than the signal of convective activity, whereas these signals were comparable for the MJO. This indicates that convective suppression in the IO contributes more to the RMM projection than to convective enhancement in the WP. Therefore, it appears that the onset of an MJO event and an RCC event are difficult to distinguish solely from the projection of atmospheric circulation and convection patterns to the RMM. Moreover, the preceding suppression detected by this composite analysis is not a sufficient condition for eastward propagation of convection nor is it necessary for initiating a burst of a large-scale convective event.

Differences in the spatial patterns of convection between the MJO and RCCs were compared by using OLR anomaly composites (Figs. 5 and 6). We compared the OLR anomaly composites of the IO-RCC and WP-RCC at day 0 with the MJO composite at day 0 and at day 15, respectively. The composite dates of the MJO were chosen to ensure that the location of the negative OLR signal approximately matched that of the RCC type at its initiation on the composite Hovmöller diagram. Comparison of the MJO and IO-RCC initiation OLR anomalies show that although both indicate large-scale convection over the IO region, the IO-RCC convective activity was weaker, and the center was shifted eastward. This could have been caused by the wider range of RMM phases that occurs with IO-RCC initiation. Correspondingly, the suppression signal also shifted eastward for the IO-RCC as compared with the MJO. On the contrary, the WP-RCC initiation was characterized by strong suppression over the IO and multiple smaller patches of convective activity over the WP. It is apparent that the WP-RCC lacks the robust and widespread negative OLR signal across the eastern edge of the MC to the WP found in the MJO. This suggests that the WP-RCC is characterized by the prevailing conditions of convective inhibition over the IO rather than by convection enhancement over the WP. It is understandable that projection of the RMM to such an atmospheric state is high in phases 5–7 owing to high sensitivity of RMM to signals over the IO. The sensitivity of RMM to convective signals over the IO arises from the spatial structure of the EOFs of RMM, in which the highest normalized magnitude for OLR occurs around 90°E in EOF2 (WH04). On the basis of these results, we restricted further analysis to the IO-RCC and MJO because we considered that comparing the WP-RCC and MJO was not expedient in investigating the properties that distinguish the MJO from other large-scale convective events.

Fig. 5.

Composite of OLR anomalies (W m−2) of the MJO at (top) day 0 and (bottom) day 15. White contours indicate where the signal is statistically significant at 90%.

Fig. 5.

Composite of OLR anomalies (W m−2) of the MJO at (top) day 0 and (bottom) day 15. White contours indicate where the signal is statistically significant at 90%.

Fig. 6.

Composite of OLR anomalies (W m−2) for (top) IO-RCC and (bottom) WP-RCC at day 0. White contours indicate where the signal is statistically significant at 90%.

Fig. 6.

Composite of OLR anomalies (W m−2) for (top) IO-RCC and (bottom) WP-RCC at day 0. White contours indicate where the signal is statistically significant at 90%.

b. MSE budget of MJO and IO-RCC

To compare the characteristics of the moisture budget that drives the circulation of MJO and IO-RCC convection, we conducted a column-integrated MSE budget analysis (e.g., Sobel et al. 2014), which can be written as

 
formula

using brackets to denote column integration, where h is the MSE; v is the velocity; LH and SH are the surface latent and sensible heat fluxes, respectively; and LW and SW are the longwave and shortwave heating rates, respectively. We divided the equatorial region (5°S–5°N) from 50°E to 160°W into three segments of 50° longitude to represent the IO (50°–100°E), MC (100°–150°E), and WP (150°E–160°W) from west to east. We compared the time evolution of the 11-day smoothed anomaly of the MSE budget term of each regional section between the MJO and IO-RCC (Figs. 7 and 8). Composites of the time evolution were centered at the initial day of the active period of the MJO and IO-RCC as in the previous analysis.

Fig. 7.

Time evolution of 11-day running-mean MSE budget terms (W m−2) during the MJO for the (left) IO, (middle) MC, and (right) WP regions. Colors signify vertical advection (red), horizontal advection (purple), net radiation (yellow), latent heat (blue), and sensible heat (orange). Day 0 was set as the day of initiation of the active phase of the MJO.

Fig. 7.

Time evolution of 11-day running-mean MSE budget terms (W m−2) during the MJO for the (left) IO, (middle) MC, and (right) WP regions. Colors signify vertical advection (red), horizontal advection (purple), net radiation (yellow), latent heat (blue), and sensible heat (orange). Day 0 was set as the day of initiation of the active phase of the MJO.

Fig. 8.

Time evolution of 11-day running-mean MSE budget terms (W m−2) during IO-RCC for the (left) IO, (middle) MC, and (right) WP regions. Colors signify vertical advection (red), horizontal advection (purple), net radiation (yellow), latent heat (blue), and sensible heat (orange). Day 0 was set as the day of initiation of IO-RCC events.

Fig. 8.

Time evolution of 11-day running-mean MSE budget terms (W m−2) during IO-RCC for the (left) IO, (middle) MC, and (right) WP regions. Colors signify vertical advection (red), horizontal advection (purple), net radiation (yellow), latent heat (blue), and sensible heat (orange). Day 0 was set as the day of initiation of IO-RCC events.

The MSE budget analysis shows that the convective activities of the MJO are organized into a system that interacts with the large-scale environment to induce cooperative energy fluxes and advective processes for convective activity. The occurrence of deep convection during MJO is indicated by the development of anomalously negative vertical advection. At its peak, vertical advection becomes the dominant term in the MSE budget equation. Large increases in radiative heating that coincide with decreases in MSE vertical advection also imply the development of clouds at high altitudes. The minimum of the vertical advection term in each region is preceded by high horizontal advection of MSE, which implies that during MJO, horizontal advection builds up the MSE eastward of the convective center. This is confirmed by the temporal evolution of the column-integrated MSE anomaly (Fig. 9a). Eastward movement of the column-integrated MSE anomaly is demonstrated by its later peak occurrence over the eastward regions, which was approximately day 5 for the IO, day 10 for the MC, and day 20 for the WP. We also note that horizontal MSE advection over the MC region was especially high where the MJO is frequently blocked, which could have contributed to the successful eastward passage of MJO convection over the MC to the WP. These results are consistent with those in previous studies that suggest the importance of horizontal MSE advection for eastward propagation of the MJO (Sobel et al. 2014; Kim et al. 2014). The rise in latent heat flux lagged the convective signal by approximately 5 days over the IO, but was nearly in phase by the time the MJO was over the WP. This result is consistent with Hendon and Salby (1994) who showed that low-level westerlies that lagged convective activity over the IO came into phase as the MJO proceeded to the WP.

Fig. 9.

Time evolution of 11-day running-mean column-integrated MSE anomaly (W m−2) for (a) MJO and (b) IO-RCC for the IO (green), MC (orange), and WP (magenta) regions. Day 0 was set as the day of initiation of the active phase of MJO and IO-RCC events.

Fig. 9.

Time evolution of 11-day running-mean column-integrated MSE anomaly (W m−2) for (a) MJO and (b) IO-RCC for the IO (green), MC (orange), and WP (magenta) regions. Day 0 was set as the day of initiation of the active phase of MJO and IO-RCC events.

Fluctuation of the MSE budget terms associated with activities of deep convection, such as the vertical advection and net radiation terms, were significantly weaker for the IO-RCC and appeared only over the IO and MC (Fig. 8). Over the IO, the peak vertical advection term was approximately half, and the peak radiative heating term was approximately one-third of the peak values during the MJO. By the time convection shifted eastward to the MC, the magnitude of these signals was even weaker. Over the WP, the signs of these signals remained consistent throughout the analysis period. That is, they were positive for the vertical advection term and negative for the net radiation term, which indicates weak convective activity. In relation to the weaker convective activity, the latent heat flux that followed the convective activity was also notably smaller. We also note that, as compared with the MJO, horizontal advection of the MSE was small prior to the convective initiation over the IO and was barely existent in the MC and WP. As a result, the IO-RCC convection initiated over conditions with a small buildup of MSE and moisture over the IO and MC, and no buildup of MSE over the WP (Fig. 9b). The characteristics of the MSE budget terms show that although IO-RCC are events of convective activity that had significant amplitude of RMM projection in phases 2–4, the nature of the convective activity is much shallower. This is indicated by the small magnitude of the vertical advection term, and the suggestion that convection is lacking large-scale circulation that strongly interacts with the large-scale environment.

Comparison of the evolution of the MSE budget terms between the MJO and IO-RCC show that the MJO develops in an environment that potentially permits MSE buildup across MJO convective regions. Conversely, the IO-RCC develops in an environment where positive MSE feedback is not favored. In the following section, we investigate the large-scale environmental conditions that lead to the contrast in the response of the MSE budget terms between the MJO and IO-RCC.

c. SST fields during MJO and RCC

We compared the SST fields associated with the MJO and IO-RCC to investigate whether the background SST fields contributed to the differences in the MSE budget terms. We believe that SST variation at the MJO and longer time scale acts as a background SST field for convective activities that influence the large-scale MSE budget through its effects on surface moisture flux and advection, and by inducing low-level convergence. With this in mind, composites of SST anomalies (SSTA) that occurred during the MJO and IO-RCC were made. The SSTA were separated into low- and high-frequency SSTA (LF-SSTA and HF-SSTA, respectively) fields to distinguish those signals at time scales longer than the MJO from those at time scales comparable with the MJO (Figs. 10 and 11). LF-SSTA is defined as a 60-day low-pass-filtered SSTA and HF-SSTA is defined as a 20–60-day bandpass-filtered SSTA using a Lanczos filter (Duchon 1979). We selected 60 days as the threshold between the LF-SSTA and HF-SSTA because MJO convection requires approximately 30 days to propagate from the IO to 180° (Fig. 4a).

Fig. 10.

(a),(b) Composite of HF-SSTA at day 0 and (c),(d) composite Hovmöller diagram of 15°S–15°N averaged HF-SSTA (C°) and OLR anomalies (W m−2) for the (left) MJO and (right) IO-RCC. Contours on the day 0 composite in (a) and (b) indicate a statistically significant signal at 90%.

Fig. 10.

(a),(b) Composite of HF-SSTA at day 0 and (c),(d) composite Hovmöller diagram of 15°S–15°N averaged HF-SSTA (C°) and OLR anomalies (W m−2) for the (left) MJO and (right) IO-RCC. Contours on the day 0 composite in (a) and (b) indicate a statistically significant signal at 90%.

Fig. 11.

(a),(b) Composite of LF-SSTA at day 0 and (c),(d) composite Hovmöller diagram of 15°S–15°N averaged LF-SSTA (C°) and OLR anomalies (W m−2) for the (left) MJO and (right) IO-RCC.Contours on the day 0 composite in (a) and (b) indicate a statistically significant signal at 90%.

Fig. 11.

(a),(b) Composite of LF-SSTA at day 0 and (c),(d) composite Hovmöller diagram of 15°S–15°N averaged LF-SSTA (C°) and OLR anomalies (W m−2) for the (left) MJO and (right) IO-RCC.Contours on the day 0 composite in (a) and (b) indicate a statistically significant signal at 90%.

Examination of the composite HF-SSTA showed that the MJO and IO-RCC were accompanied by similar response in the SST to the preceding suppression and to the following convective activities (Figs. 10a and 10b). The HF-SSTA pattern was quite similar for both the MJO and IO-RCC. After initiation, at day 0, both the MJO and IO-RCC showed a common feature of negative SSTA spreading over the IO and a positive anomaly prevailing over the WP (Figs. 10c and 10d). However, the eastern edge of the MC at approximately 120°E, showed an opposite HF-SSTA sign between the MJO and IO-RCC, with the MJO having a positive SSTA. This is consistent with our finding that the center of convection shifted eastward for IO-RCC as compared with the MJO; the signal of negative OLR of the IO-RCC extended to the eastern edge of the MC, where the opposite SSTA for the MJO was located (Fig. 6). A similar SST response to convective activity was apparent in the composite Hovmöller diagram for the HF-SSTAs of MJO and IO-RCC. Both showed increased SSTAs with the passage of the suppressed convective conditions from the IO to the WP and decreases in the anomaly as convection developed over the IO. Such responses by the ocean to the atmosphere are consistent with the ocean–atmospheric feedback documented in previous studies (Shinoda et al. 1998; DeMott et al. 2015). Although the magnitude of the SSTA fluctuation was larger for the MJO, owing to its stronger and longer-lasting convection, the overall pattern of HF-SSTA preceding and during convection over the IO was very similar to that of the IO-RCC. These results imply that the coupling of convection and the ocean is similar between MJO and IO-RCC. Therefore, it follows that regarding the high-frequency response, convective activity and suppression of the MJO and IO-RCC have similar effects on SSTs.

However, the MJO and IO-RCC are associated with strikingly different LF-SSTA fields (Fig. 11). At their initiation, the MJO and IO-RCC displayed a nearly reversed LF-SSTA pattern from the eastern MC to the WP (Figs. 11a and 11b). A warm anomaly spread from the equatorial WP for the MJO, whereas a cold anomaly occurs for the IO-RCC. This contrast was the most distinct over the WP, where the positive and negative anomalies exhibited respective peaks for the MJO and IO-RCC. Although it was not as strong as over the WP, the MJO (IO-RCC) shows a positive (negative) LF-SSTA pattern over the IO as well. Moreover, both the MJO and IO-RCC showed a positive LF-SSTA field over the MC region. We also note that at the initiation of the MJO, an overall increase in the equatorial LF-SSTA occurred toward its peak over the WP. The different LF-SSTA patterns for the MJO and IO-RCC were also distinct in the Hovmöller composite of LF-SSTA (Figs. 11c and 11d). For the MJO, positive LF-SSTA began to prevail across the entire region from the IO to WP from approximately 10 days prior to initiation until the passage of the convective activity. In contrast, for the IO-RCC, the LF-SSTA was positive over the MC and negative elsewhere for almost the entire event duration, with the exception of a warm anomaly over the western IO before initiation. It is apparent that IO-RCC convective activities initiated over the IO, where the LF-SSTA is relatively warm at approximately 60°E, and spread eastward only up to the point at which the LF-SSTA reached its maximum at approximately 110°E. This result implies that although the MJO developed under conditions in which the buildup of moisture and MSE was supported by the eastward-increasing LF-SSTA pattern, a supply of moisture and buildup of MSE sufficient for development of convective activity into a large-scale system was not supported eastward of the MC during the IO-RCC owing to the negative LF-SSTA over the WP.

From the results of the LF-SSTA analysis, we speculate that background conditions with an eastward increase of SST from the IO to the WP are important for development of MJO. We analyze the frequency of MJO events as a function of SST difference between the equatorial WP and IO to provide statistical evidence for the importance of the zonal SST gradient on the MJO. The WP and IO regions are defined as the area within 5°S–5°N, 120°E–180°, and 5°S–5°N, 60°–120°E, respectively, and the difference in area-averaged SST between the WP and IO is defined as a simple index of zonal SST gradient (WP-IO). The frequency distribution of daily WP-IO during 1982–2012 binned by intervals of 0.4°C is shown in Fig. 12. The WP-IO peak was centered at 0.2°C, which is consistent with the climatological average of WP-IO at 0.3°C. With this in mind, histograms of the ratio of occurrence of MJO and IO-RCC sorted by the WP-IO value averaged through the lifetime of each event are shown in Figs. 13a and 13b. It is apparent that, although MJO events were distributed around the mean WP-IO value with a slightly left skew, the IO-RCC occurrences were concentrated at the mean. Considering the basic WP-IO frequency distribution, we normalized the number of MJO and IO-RCC events in each WP-IO bin by the number of days of WP-IO itself in each bin from 1982 to 2012. This shows the deviation of the MJO and IO-RCC frequency distribution from the frequency distribution of WP-IO. This normalized frequency distribution helped to clarify the actual influence of WP-IO on MJO and IO-RCC development. The normalized MJO frequency (Fig. 13c and 13d) increased with the WP-IO value, which indicated that the MJO tends to be more frequent when the WP-IO is higher. Conversely, the normalized frequency of the IO-RCC showed no consistent trend with the WP-IO and assumed a similar value across the range of the WP-IO.

Fig. 12.

Histogram of area-averaged SST difference of WP (5°S–5°N, 120°E–180°) from IO (5°S–5°N, 60°–120°E) in 1982–2012. Bins are at 0.4°C intervals with median values indicated at the bottom.

Fig. 12.

Histogram of area-averaged SST difference of WP (5°S–5°N, 120°E–180°) from IO (5°S–5°N, 60°–120°E) in 1982–2012. Bins are at 0.4°C intervals with median values indicated at the bottom.

Fig. 13.

Ratio of occurrence of (a),(c) MJO and (b),(d) IO-RCC (top) sorted by the mean WP SST–IO SST during the event and (bottom) normalized ratio by the total number of occurrences of WP-IO values between 1982 and 2012.

Fig. 13.

Ratio of occurrence of (a),(c) MJO and (b),(d) IO-RCC (top) sorted by the mean WP SST–IO SST during the event and (bottom) normalized ratio by the total number of occurrences of WP-IO values between 1982 and 2012.

It is conceivable that such a large-scale SST pattern is linked to interannual variability of the El Niño–Southern Oscillation (ENSO; Trenberth 1997) and El Niño Modoki (Weng et al. 2007). Regarding the ENSO, there was no significant bias noted in the frequency of occurrence in the MJO or IO-RCC. Figure 14 shows the ratio of occurrence of MJO and IO-RCC for a given range of SSTA of the Niño-3.4 index. For both the MJO and IO-RCC, the peak was located around the center of the distribution of the Niño-3.4 index. The unevenness of the IO-RCC histogram is most likely attributed to the small sample size. Furthermore, the patterns of composite LF-SST for both the MJO and IO-RCC were statistically insignificant eastward of 150°W (not shown), which indicates that the eastern Pacific SST is not dominant in modulating MJO or IO-RCC activities. In contrast, the MJO frequency of occurrence appears to be related to El Niño Modoki, which is associated with SSTAs over the central Pacific. The ratio of occurrence of MJO and IO-RCC for a given range of El Niño Modoki index (EMI; Ashok et al. 2007) are shown in Fig. 15 where EMI is defined as difference between the area-averaged SSTA in the central Pacific (10°S–10°N, 165°E–140°W), and that in the eastern and WP (15°S–5°N, 110°–70°W and 10°S–20°N, 125°–145°E, respectively). The figure indicates that although the MJO tends to occur more frequently when the EMI is high, the occurrence of IO-RCC is apparently unaffected by the EMI. This result is consistent with the LF-SSTA analysis showing that positive SSTAs from the WP to central Pacific are favorable for MJO events.

Fig. 14.

Ratio of (a) MJO and (b) IO-RCC occurrence according to the Niño-3.4 index value at their initiation. Bins are at 1.0°C intervals with median values indicated at the bottom.

Fig. 14.

Ratio of (a) MJO and (b) IO-RCC occurrence according to the Niño-3.4 index value at their initiation. Bins are at 1.0°C intervals with median values indicated at the bottom.

Fig. 15.

Ratio of (left) MJO and (right) IO-RCC occurrence according to the El Niño Modoki index value at their initiation.

Fig. 15.

Ratio of (left) MJO and (right) IO-RCC occurrence according to the El Niño Modoki index value at their initiation.

Our analysis also confirms that the SST pattern associated with the MJO, characterized by a spreading warm anomaly over the WP, is present in all seasons. A composite of the LF-SSTA for each season was made by categorizing MJO events according to their initiation date: December–February (DJF), March–May (MAM), June–August (JJA), and September–November (SON; Fig. 16). Seasonal differences were apparent, especially over the eastern edge of the MC, where LF-SSTAs are negative for DJF and MAM, but positive for JJA and SON. However, we note that all seasons showed positive LF-SSTAs over the equatorial WP from 150°E to 180°, where positive LF-SSTAs were found to be significant in an MJO composite of all events. Although some seasonal variations were noted, the overall results reinforce our view that high SST over the WP enhances MJO development.

Fig. 16.

Composite of LF-SSTA at initiation of MJO during each season: (a) DJF, (b) MAM, (c) JJA, and (d) SON.

Fig. 16.

Composite of LF-SSTA at initiation of MJO during each season: (a) DJF, (b) MAM, (c) JJA, and (d) SON.

5. Summary and discussion

This study revealed MJO characteristics that distinguish them from tentatively organized convective events that fail to become an MJO. To identify these characteristics, we classified days with a high RMM amplitude into MJO and two types of RCC events by incorporating continuity into the RMM phase sequence. Two types of RCC were categorized according to the location of the convective activity, in which IO-RCC and WP-RCC were associated with convective activity over the IO and WP, respectively. However, it was determined that WP-RCC was characterized by strong suppression over the IO and only weak convective activity over the WP. Therefore, WP-RCC events were not included in further analysis and comparisons between the MJO and IO-RCC were conducted. Although this simple classification method for MJO and RCC is by no means definitive, our findings demonstrated that the results are sufficient for distinguishing intraseasonal eastward-propagating events as MJO and short-lived stagnant convective events as RCC.

Intriguingly, comparison of the convectively suppressed state associated with the MJO and IO-RCC indicated that the intensity of the convective suppression preceding a convective event was not a critical factor in determining whether a convective event becomes an MJO. However, the MSE anomaly that developed in association with the circulation differed; convective activities of the MJO were preceded by significant buildup of MSE and were accompanied by cooperative MSE flux and advective processes for convection. Such MSE accumulation processes were weak during the IO-RCC. Therefore, we infer that initiation of convection over an environment with the potential for MSE buildup is important for the development of an MJO. Thus, we investigated for such an MJO enhancing background through analysis of the SST.

Analysis of different frequency SSTA fields showed that the LF-SSTA pattern distinguished the MJO from IO-RCC. Composites of the LF-SSTA showed contrasting patterns for the MJO and IO-RCC, in which they were associated with positive and negative SSTAs over the WP, respectively. Conversely, the HF-SSTA pattern, arising from changes in SST owing to air–sea interaction, showed a similar spatial and temporal evolution pattern for both the MJO and IO-RCC. Our results showed that as a function of a positive zonal SST gradient from IO to WP, likelihood of MJO occurrence rises with an increase in the zonal SST gradient. Hence, we infer that warming of the SST over the WP and generating a positive zonal SST gradient from the IO to the WP at time scales longer than intraseasonal is important for MJO development.

On the basis of our findings, global climate models (GCMs) in the absence of air–sea coupling should be able to reproduce the MJO as long as the prescribed SST is favorable for MJO development. This inference is supported by the fact that some GCMs are capable of simulating MJO-like disturbances if observed or climatological mean SSTs are given and that their representations are improved by air–sea coupling (e.g., Inness and Slingo 2003; Benedict and Randall 2011; Klingaman and Woolnough 2014; DeMott et al. 2014). Furthermore, the basic-state SST field that we consider to be essential is on a time scale much longer than that of MJO and is not a product of SST response to convective suppression (Shinoda et al. 1998). However, changes in the HF-SSTA from air–sea coupling can supplement the MJO supportive background SSTs within an MJO time scale, as reported by Flatau et al. (1997) and Hirata et al. (2013). Therefore, although we assert that the influence of LF-SSTA is dominant over that of HF-SSTA for MJO development, our findings do not conflict with their work or others (e.g., Hendon et al. 1999; Klingaman and Woolnough 2014).

Slowly changing SST is an influential environmental factor for the MJO because it can provide a near-standing bottom boundary condition for the atmosphere on an intraseasonal time scale. For example, episodes of interannual variability such as positive ENSO and positive IO dipole (IOD) modes tend to produce faster MJO events (Izumo et al. 2010; Pohl and Matthews 2007) and vice versa. Prior research has indicated that MJO events occur more frequently, and are enhanced during negative IOD events (Ling et al. 2013; Shinoda and Han 2005; Izumo et al. 2010) and after La Niña events (Pohl and Matthews 2007). Complementary to these studies, we propose in the present study that background SSTs with a positive zonal SST gradient toward the WP from the IO play a fundamental role in developing the MJO by inducing low-level equatorial westerlies, as shown by the one-layer model of Lindzen and Nigam (1987). Observational evidence that MJO activity peaks with the season and location of mean low-level westerlies (Zhang and Dong 2004), and model dependency on simulating correct mean westerly winds for simulating MJO (Inness et al. 2003) also corroborate with our results. It should also be noted that the overall positive LF-SSTA across the IO to the WP associated with the MJO contributes to a large-scale potential for the necessary buildup of MSE for maintaining active convection on an intraseasonal time scale. The positive zonal SST gradient toward the WP also contributes to MJO propagation in that it enables greater MSE buildup on the eastern regions.

However, it is important to remember, that we have solely identified an environmental background favorable for MJO. The degree to which it controls the development of the MJO still needs to be assessed. Moreover, the realization and maintenance of such a long-term environmental background favorable for MJO development should be studied in future work. Such factors are essential for a comprehensive understanding of the MJO.

Acknowledgments

The authors wish to express their thanks to the three anonymous reviewers for providing careful and pertinent comments on the originally submitted manuscript. This study is supported by Grant-in-Aid for scientific research from the Japan Society for the Promotion of Science (Grants 16J07769, B-25287119, and B-16H04048).

REFERENCES

REFERENCES
Ashok
,
K.
,
S. K.
Behera
,
S. A.
Rao
,
H.
Weng
, and
T.
Yamagata
,
2007
:
El Niño Modoki and its possible teleconnection
.
J. Geophys. Res.
,
112
,
C11007
, https://doi.org/10.1029/2006JC003798.
Australian Bureau of Meteorology
,
2018
: Madden–Julian oscillation: RMM data (updated daily). Commonwealth of Australia, Bureau of Meteorology, accessed 4 April 2018, http://www.bom.gov.au/climate/mjo/graphics/rmm.74toRealtime.txt.
Benedict
,
J. J.
, and
D. A.
Randall
,
2011
:
Impacts of idealized air–sea coupling on Madden–Julian oscillation structure in the superparameterized CAM
.
J. Atmos. Sci.
,
68
,
1990
2008
, https://doi.org/10.1175/JAS-D-11-04.1.
Bretherton
,
C. S.
,
M. E.
Peters
, and
L. E.
Back
,
2004
:
Relationships between water vapor path and precipitation over the tropical oceans
.
J. Climate
,
17
,
1517
1528
, https://doi.org/10.1175/1520-0442(2004)017<1517:RBWVPA>2.0.CO;2.
DeMott
,
C. A.
,
C.
Stan
,
D. A.
Randall
, and
M. D.
Branson
,
2014
:
Intraseasonal variability in coupled GCMs: The roles of ocean feedbacks and model physics
.
J. Climate
,
27
,
4970
4995
, https://doi.org/10.1175/JCLI-D-13-00760.1.
DeMott
,
C. A.
,
N. P.
Klingaman
, and
S. J.
Woolnough
,
2015
:
Atmosphere–ocean coupled processes in the Madden–Julian oscillation
.
Rev. Geophys.
,
53
,
1099
1154
, https://doi.org/10.1002/2014RG000478.
Duchon
,
C. E.
,
1979
:
Lanczos filtering in one and two dimensions
.
J. Appl. Meteor.
,
18
,
1016
1022
, https://doi.org/10.1175/1520-0450(1979)018<1016:LFIOAT>2.0.CO;2.
Feng
,
J.
,
T.
Li
, and
W.
Zhu
,
2015
:
Propagating and nonpropagating MJO events over Maritime Continent
.
J. Climate
,
28
,
8430
8449
, https://doi.org/10.1175/JCLI-D-15-0085.1.
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
, https://doi.org/10.1175/1520-0469(1997)054<2373:TFBECA>2.0.CO;2.
Hendon
,
H. H.
, and
M. L.
Salby
,
1994
:
The life cycle of the Madden–Julian oscillation
.
J. Atmos. Sci.
,
51
,
2225
2237
, https://doi.org/10.1175/1520-0469(1994)051<2225:TLCOTM>2.0.CO;2.
Hendon
,
H. H.
,
C.
Zhang
, and
J. D.
Glick
,
1999
:
Interannual variation of the Madden–Julian oscillation during austral summer
.
J. Climate
,
12
,
2538
2550
, https://doi.org/10.1175/1520-0442(1999)012<2538:IVOTMJ>2.0.CO;2.
Hirata
,
F. E.
,
P. J.
Webster
, and
V. E.
Toma
,
2013
:
Distinct manifestations of austral summer tropical intraseasonal oscillations
.
Geophys. Res. Lett.
,
40
,
3337
3341
, https://doi.org/10.1002/grl.50632.
Inness
,
P. M.
, and
J. M.
Slingo
,
2003
:
Simulation of the Madden–Julian oscillation in a coupled general circulation model. Part I: Comparison with observations and an atmosphere-only GCM
.
J. Climate
,
16
,
345
364
, https://doi.org/10.1175/1520-0442(2003)016<0345:SOTMJO>2.0.CO;2.
Inness
,
P. M.
,
J. M.
Slingo
,
E.
Guilyardi
, and
J.
Cole
,
2003
:
Simulation of the Madden–Julian oscillation in a coupled general circulation model. Part II: The role of the basic state
.
J. Climate
,
16
,
365
382
, https://doi.org/10.1175/1520-0442(2003)016<0365:SOTMJO>2.0.CO;2.
Izumo
,
T.
,
S.
Masson
,
J.
Vialard
,
C.
de Boyer Montegut
,
S. K.
Behera
,
G.
Madec
,
K.
Takahashi
, and
T.
Yamagata
,
2010
:
Low and high frequency Madden–Julian oscillations in austral summer: Interannual variations
.
Climate Dyn.
,
35
,
669
683
, https://doi.org/10.1007/s00382-009-0655-z.
Kalnay
,
E.
, and Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kiladis
,
G. N.
,
J.
Dias
,
K. H.
Straub
,
M. C.
Wheeler
,
S. N.
Tulich
,
K.
Kikuchi
,
K. M.
Weickmann
, and
M. J.
Ventrice
,
2014
:
A comparison of OLR and circulation-based indices for tracking the MJO
.
Mon. Wea. Rev.
,
142
,
1697
1715
, https://doi.org/10.1175/MWR-D-13-00301.1.
Kim
,
D.
,
J. S.
Kug
, and
A. H.
Sobel
,
2014
:
Propagating versus nonpropagating Madden-Julian oscillation events
.
J. Climate
,
27
,
111
125
, https://doi.org/10.1175/JCLI-D-13-00084.1.
Klingaman
,
N. P.
, and
S. J.
Woolnough
,
2014
:
The role of air-sea coupling in the simulation of the Madden-Julian oscillation in the Hadley Centre model
.
Quart. J. Roy. Meteor. Soc.
,
140
,
2272
2286
, https://doi.org/10.1002/qj.2295.
Liebmann
,
B.
, and
C.
Smith
,
1996
:
Description of a complete (interpolated) outgoing longwave radiation dataset
.
Bull. Amer. Meteor. Soc.
,
77
,
1275
1277
, https://doi.org/10.1175/1520-0477-77.6.1274.
Lindzen
,
R. S.
, and
S.
Nigam
,
1987
:
On the role of sea surface temperature gradients in forcing low-level winds and convergence in the tropics
.
J. Atmos. Sci.
,
44
,
2418
2436
, https://doi.org/10.1175/1520-0469(1987)044<2418:OTROSS>2.0.CO;2.
Ling
,
J.
,
C.
Zhang
, and
P.
Bechtold
,
2013
:
Large-scale distinctions between MJO and non-MJO convective initiation over the tropical Indian Ocean
.
J. Atmos. Sci.
,
70
,
2696
2712
, https://doi.org/10.1175/JAS-D-13-029.1.
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
, https://doi.org/10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.
Majda
,
A. J.
, and
S. N.
Stechmann
,
2009
:
The skeleton of tropical intraseasonal oscillations
.
Proc. Natl. Acad. Sci. USA
,
106
,
8417
8422
, https://doi.org/10.1073/pnas.0903367106.
Matthews
,
A. J.
,
2008
:
Primary and successive events in the Madden–Julian oscillation
.
Quart. J. Roy. Meteor. Soc.
,
134
,
439
453
, https://doi.org/10.1002/qj.224.
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
, https://doi.org/10.1126/science.1148443.
Miura
,
H.
,
T.
Suematsu
, and
T.
Nasuno
,
2015
:
An ensemble hindcast of the Madden-Julian oscillation during the CINDY2011/DYNAMO field campaign and influence of seasonal variation of sea surface temperature
.
J. Meteor. Soc. Japan
,
93A
,
115
137
, https://doi.org/10.2151/jmsj.2015-055.
Pohl
,
B.
, and
A. J.
Matthews
,
2007
:
Observed changes in the lifetime and amplitude of the Madden–Julian Oscillation associated with interannual ENSO sea surface temperature anomalies
.
J. Climate
,
20
,
2659
2674
, https://doi.org/10.1175/JCLI4230.1.
Reynolds
,
R. W.
,
N. A.
Rayner
,
T. M.
Smith
,
D. C.
Stokes
, and
W.
Wang
,
2002
:
An improved in situ and satellite SST analysis for climate
.
J. Climate
,
15
,
1609
1625
, https://doi.org/10.1175/1520-0442(2002)015<1609:AIISAS>2.0.CO;2.
Rui
,
H.
, and
B.
Wang
,
1990
:
Development characteristics and dynamic structure of tropical intraseasonal convection anomalies
.
J. Atmos. Sci.
,
47
,
357
379
, https://doi.org/10.1175/1520-0469(1990)047<0357:DCADSO>2.0.CO;2.
Shinoda
,
T.
, and
W.
Han
,
2005
:
Influence of the Indian Ocean dipole on atmospheric subseasonal variability
.
J. Climate
,
18
,
3891
3909
, https://doi.org/10.1175/JCLI3510.1.
Shinoda
,
T.
,
H. H.
Hendon
, and
J.
Glick
,
1998
:
Intraseasonal variability of surface fluxes and sea surface temperature in the tropical western Pacific and Indian Oceans
.
J. Climate
,
11
,
1685
1702
, https://doi.org/10.1175/1520-0442(1998)011<1685:IVOSFA>2.0.CO;2.
Sobel
,
A.
,
S.
Wang
, and
D.
Kim
,
2014
:
Moist static energy budget of the MJO during DYNAMO
.
J. Atmos. Sci.
,
71
,
4276
4291
, https://doi.org/10.1175/JAS-D-14-0052.1.
Stachnik
,
J. P.
,
D. E.
Waliser
, and
A. J.
Majda
,
2015
:
Precursor environmental conditions associated with the termination of Madden–Julian oscillation events
.
J. Atmos. Sci.
,
72
,
1908
1931
, https://doi.org/10.1175/JAS-D-14-0254.1.
Straub
,
K. H.
,
2013
:
MJO initiation in the real-time multivariate MJO index
.
J. Climate
,
26
,
1130
1151
, https://doi.org/10.1175/JCLI-D-12-00074.1.
Trenberth
,
K. E.
,
1997
:
The definition of El Niño
.
Bull. Amer. Meteor. Soc.
,
78
,
2771
2777
, https://doi.org/10.1175/1520-0477(1997)078<2771:TDOENO>2.0.CO;2.
Weng
,
H.
,
K.
Ashok
,
S. K.
Behera
,
S. A.
Rao
, and
T.
Yamagata
,
2007
:
Impacts of recent El Niño Modoki on dry/wet conditions in the Pacific rim during boreal summer
.
Climate Dyn.
,
29
,
113
129
, https://doi.org/10.1007/s00382-007-0234-0.
Wheeler
,
M.
, and
K. M.
Weickmann
,
2001
:
Real-time monitoring and prediction of modes of coherent synoptic to intraseasonal tropical variability
.
Mon. Wea. Rev.
,
129
,
2677
2694
, https://doi.org/10.1175/1520-0493(2001)129<2677:RTMAPO>2.0.CO;2.
Wheeler
,
M.
, and
H. H.
Hendon
,
2004
:
An all-season real-time multivariate MJO index: Development of an index for monitoring and prediction
.
Mon. Wea. Rev.
,
132
,
1917
1932
, https://doi.org/10.1175/1520-0493(2004)132<1917:AARMMI>2.0.CO;2.
Zhang
,
C.
,
2005
:
Madden-Julian oscillation
.
Rev. Geophys.
,
43
,
RG2003
, https://doi.org/10.1029/2004RG000158.
Zhang
,
C.
,
2013
:
Madden–Julian oscillation: Bridging weather and climate
.
Bull. Amer. Meteor. Soc.
,
94
,
1849
1870
, https://doi.org/10.1175/BAMS-D-12-00026.1.
Zhang
,
C.
, and
M.
Dong
,
2004
:
Seasonality in the Madden–Julian oscillation
.
J. Climate
,
17
,
3169
3180
, https://doi.org/10.1175/1520-0442(2004)017<3169:SITMO>2.0.CO;2.
Zhang
,
C.
, and
J.
Ling
,
2017
:
Barrier effect of the Indo-Pacific Maritime Continent on the MJO: Perspectives from tracking MJO precipitation
.
J. Climate
,
30
,
3439
3459
, https://doi.org/10.1175/JCLI-D-16-0614.1.
Zhu
,
J.
,
W.
Wang
, and
A.
Kumar
,
2017
:
Simulations of MJO propagation across the Maritime Continent: Impacts of SST feedback
.
J. Climate
,
30
,
1689
1704
, https://doi.org/10.1175/JCLI-D-16-0367.1.

Footnotes

© 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).