Influence of the Stratospheric Quasi-Biennial Oscillation on the Madden–Julian Oscillation during Austral Summer

Eriko Nishimoto Department of Geophysics, Kyoto University, Kyoto, Japan

Search for other papers by Eriko Nishimoto in
Current site
Google Scholar
PubMed
Close
and
Shigeo Yoden Department of Geophysics, Kyoto University, Kyoto, Japan

Search for other papers by Shigeo Yoden in
Current site
Google Scholar
PubMed
Close
Full access

We are aware of a technical issue preventing figures and tables from showing in some newly published articles in the full-text HTML view.
While we are resolving the problem, please use the online PDF version of these articles to view figures and tables.

Abstract

Influence of the stratospheric quasi-biennial oscillation (QBO) on the Madden–Julian oscillation (MJO) and its statistical significance are examined for austral summer (DJF) in neutral ENSO events during 1979–2013. The amplitude of the OLR-based MJO index (OMI) is typically larger in the easterly phase of the QBO at 50 hPa (E-QBO phase) than in the westerly (W-QBO) phase. Daily composite analyses are performed by focusing on phase 4 of the OMI, when the active convective system is located over the eastern Indian Ocean through the Maritime Continent. The composite OLR anomaly shows a larger negative value and slower eastward propagation with a prolonged period of active convection in the E-QBO phase than in the W-QBO phase. Statistically significant differences of the MJO activities between the QBO phases also exist with dynamical consistency in the divergence of horizontal wind, the vertical wind, the moisture, the precipitation, and the 100-hPa temperature. A conditional sampling analysis is also performed by focusing on the most active convective region for each day, irrespective of the MJO amplitude and phase. Composite vertical profiles of the conditionally sampled data over the most active convective region reveal lower temperature and static stability around the tropopause in the E-QBO phase than in the W-QBO phase, which indicates more favorable conditions for developing deep convection. This feature is more prominent and extends into lower levels in the upper troposphere over the most active convective region than other tropical regions. Composite longitude–height sections show similar features of the large-scale convective system associated with the MJO, including a vertically propagating Kelvin response.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author e-mail: Eriko Nishimoto, eriko@kugi.kyoto-u.ac.jp

Abstract

Influence of the stratospheric quasi-biennial oscillation (QBO) on the Madden–Julian oscillation (MJO) and its statistical significance are examined for austral summer (DJF) in neutral ENSO events during 1979–2013. The amplitude of the OLR-based MJO index (OMI) is typically larger in the easterly phase of the QBO at 50 hPa (E-QBO phase) than in the westerly (W-QBO) phase. Daily composite analyses are performed by focusing on phase 4 of the OMI, when the active convective system is located over the eastern Indian Ocean through the Maritime Continent. The composite OLR anomaly shows a larger negative value and slower eastward propagation with a prolonged period of active convection in the E-QBO phase than in the W-QBO phase. Statistically significant differences of the MJO activities between the QBO phases also exist with dynamical consistency in the divergence of horizontal wind, the vertical wind, the moisture, the precipitation, and the 100-hPa temperature. A conditional sampling analysis is also performed by focusing on the most active convective region for each day, irrespective of the MJO amplitude and phase. Composite vertical profiles of the conditionally sampled data over the most active convective region reveal lower temperature and static stability around the tropopause in the E-QBO phase than in the W-QBO phase, which indicates more favorable conditions for developing deep convection. This feature is more prominent and extends into lower levels in the upper troposphere over the most active convective region than other tropical regions. Composite longitude–height sections show similar features of the large-scale convective system associated with the MJO, including a vertically propagating Kelvin response.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author e-mail: Eriko Nishimoto, eriko@kugi.kyoto-u.ac.jp

1. Introduction

An understanding of the stratosphere–troposphere dynamical coupling in the extratropics has been deepened for the last decade or two through observational data analyses and numerical model experiments (e.g., Baldwin and Dunkerton 2001; Yoden et al. 2002). In the tropics, on the other hand, the stratosphere–troposphere dynamical coupling (i.e., two-way interactions between them) has not received much attention, even though upward influence from the troposphere to the stratosphere has been understood rather well by the studies on several kinds of tropical waves generated in the troposphere and propagating upward into the stratosphere, ranging from planetary-scale equatorial waves to small-scale gravity waves. Such large differences in the research activities might be attributable to the difference in the fundamental dynamics between the tropics and the extratropics: in the tropics, Coriolis parameter is so small that quasigeostrophic constraint is not very strong, and mesoscale moist convection is the predominant source driving the atmospheric motions. Recently, several studies have shown the influence of some stratospheric phenomena on tropical deep convection and its organizations, including the influence of stratospheric sudden warming (Eguchi and Kodera 2010; Kodera et al. 2011; Albers et al. 2016), the stratospheric quasi-biennial oscillation (QBO; Collimore et al. 1998, 2003; Giorgetta et al. 1999; Huang et al. 2012) and its dynamical analog simulated in idealized numerical experiments (Yoden et al. 2014; Nie and Sobel 2015; Nishimoto et al. 2016), and the cooling trend of the tropical tropopause layer (Emanuel et al. 2013).

The stratospheric QBO is a dominant interannual variation in the equatorial stratosphere and is characterized as periodic variations of the easterly and westerly zonal-mean zonal wind with the period of about 2 yr (e.g., Baldwin et al. 2001). There have been several observational analyses about the relationship between the QBO and tropical deep convection. Collimore et al. (1998, 2003) showed that the interannual variations of the convective activities in the chronically convective regions are significantly correlated with the QBO-related variations of the zonal-mean zonal wind in the lower stratosphere. Liess and Geller (2012) conducted careful separation of the QBO signal from El Niño–Southern Oscillation (ENSO) and other signals and showed statistically significant increases of tropical deep convection over the western and eastern Pacific when the zonal-mean zonal wind in the lower stratosphere is easterly phase of the QBO (E-QBO phase), compared with those when it is westerly phase of the QBO (W-QBO phase).

Kuma (1990) performed spectral analysis using the zonal wind data observed by the radiosonde at Singapore and showed that the intensity of the intraseasonal oscillation at 150 hPa is well correlated with the stratospheric QBO. Liu et al. (2014) showed that amplitude of the Madden–Julian oscillation (MJO) in outgoing longwave radiation (OLR) is larger in the E-QBO phase than in the W-QBO phase during December–February (DJF). Very recently, Yoo and Son (2016) examined the relationship between the MJO and the QBO for every season and showed significantly high correlation only during austral summer (DJF) and the extended austral summer season (November–March). They evaluated the MJO amplitude by various metrics and found that the MJO amplitude during austral summer is generally larger in the E-QBO phase than in the W-QBO phase.

Dynamic and thermodynamic conditions around the tropical tropopause (~100 hPa) can be modulated with the QBO. The vertical shear of zonal wind associated with the QBO accompanies a temperature anomaly around the equator in the lower stratosphere in order to maintain thermal wind balance, and this temperature anomaly entails large-scale vertical motion via the thermodynamic budget (e.g., Plumb and Bell 1982; Baldwin et al. 2001). Therefore, the tropical tropopause is colder and located higher with an ascending anomaly in the E-QBO phase, whereas it is warmer and located lower with a descending anomaly in the W-QBO phase (Fueglistaler et al. 2009). In association with these modulations, the vertical shear of zonal wind and static stability around the tropical tropopause are also varied by the QBO. One of them or a combination of them is thought to modulate the tropical deep convection (Gray et al. 1992; Collimore et al. 2003).

Motivated by the previous work of Yoo and Son (2016), this study conducts a further data analysis on the influence of the stratospheric QBO on the MJO during austral summer (DJF) and its statistical significance. Two different types of composite analyses are performed in three QBO phases: that is, E-QBO, neutral QBO (N-QBO), and W-QBO phases. One is daily composite analyses to investigate the time evolution of the MJO by focusing on a particular phase of the MJO. The other is conditional sampling analyses focusing on the most active convective region for each day, irrespective of the amplitude and phase of the MJO.

The remainder of the paper is organized as follows. Section 2 describes indices of QBO, ENSO, and MJO and other datasets used in this study. Section 3 gives the results. First, time variations of the indices are described in section 3a. The results of daily composite analyses of OLR are described in section 3b, and those of other variables related to the evolution of the dynamic and thermodynamic structures of the MJO are in section 3c. The results of conditional sampling analyses are given in section 3d. Discussion is in section 4, and conclusions are in section 5.

2. Data

a. QBO index

The QBO phase is determined by using the European Centre for Medium-Range Weather Forecasts (ECMWF) global interim reanalysis (ERA-Interim) data (Dee et al. 2011), for 35 yr from January 1979 to December 2013. We use the monthly mean zonal-mean zonal wind at 50 hPa averaged over 10°S–10°N (U50) as a QBO index. Figure 1a shows the time series of the QBO index. The E-QBO phase is defined as the phase when U50 is less than or equal to one-half of the standard deviation below the time mean: . The W-QBO phase is defined when U50 is greater than or equal to one-half of the standard deviation above the time mean: . The rest of the period is referred to the N-QBO phase. The thresholds that separate the QBO phases are drawn with dashed lines in Fig. 1a. Months in the E-QBO, N-QBO, and W-QBO phases during austral summer (DJF) are marked with blue, green, and red symbols, respectively, in the figure. Figure 2a shows a histogram of the QBO index during austral summer. The QBO index has a trimodal distribution with peaks at around −19, −1, and 7 m s−1, and the thresholds separating the QBO phases are located around the dips between the peaks.

Fig. 1.
Fig. 1.

Time series of (a) U50, (b) Niño-3.4 SSTA, and (c) OMI amplitude for January 1979–December 2013. Dashed lines in (a) and (b) show the thresholds separating the QBO phases and ENSO phases, respectively. DJF months are shaded in gray. DJF months in the E-QBO, N-QBO, and W-QBO phases are marked by blue, green, and red color symbols, respectively, and those in the El Niño, neutral ENSO, and La Niña periods are marked with crosses, circles, and triangles, respectively.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

Fig. 2.
Fig. 2.

Histograms of (a) U50 and (c) Niño-3.4 SSTA and (b) scatterplot of U50 vs Niño-3.4 SSTA for DJF months. Stars mark the months when the MJO events are identified, whereas diamonds mark the months without any MJO event. Gray lines show the thresholds separating ENSO or QBO phases. At each phase, the number of months included in each phase is indicated, and the number of months with MJO events is given in square brackets.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

b. ENSO index

The ENSO index used in this study is based on the sea surface temperature derived from the Hadley Centre Global Sea Ice and Sea Surface Temperature (HadISST), version 1.1 (Rayner et al. 2003). The index is given by the sea surface temperature anomaly (SSTA) from the monthly climatology in Niño-3.4 region (5°S–5°N, 170°–120°W). Figure 1b shows the time series of the ENSO index. The El Niño phase is defined when Niño-3.4 SSTA is greater than 1.0 K, and the La Niña phase is when Niño-3.4 SSTA is less than −1.0 K. These threshold values are drawn with dashed lines in Fig. 1b. The neutral ENSO phase is the rest of the period. Austral summer months (DJF) in El Niño, neutral ENSO, and La Niña phases are marked with crosses, circles, and triangles, respectively, in Figs. 1a and 1b. Figure 2c shows a histogram of the ENSO index during DJF. The index is mostly distributed (about 65%) within the range between −1.0 and 1.0 K, which corresponds to the range of the neutral ENSO.

c. MJO index

The MJO index used in this study is the one based on OLR, which focuses on the large-scale convective component of the MJO (Kiladis et al. 2014). This OLR-based MJO index (OMI) is available at http://www.esrl.noaa.gov/psd/mjo/mjoindex/. Figure 1c shows a time series of the monthly averaged OMI amplitude. Circles mark DJF months in the neutral ENSO phase, and the colors of the marks show the QBO phases as noted above. The OMI consists of the first two principal components [PC1(t) and PC2(t)], and they are derived by projecting 20–96-day filtered OLR in the equatorial region within ±20° in latitudes onto the daily spatial empirical orthogonal function (EOF) of 30–96-day eastward filtered OLR (Kiladis et al. 2014). The monthly averaged OMI amplitude is calculated by averaging daily OMI amplitude . PC1 is normalized to have a standard deviation of 1.0 whereas PC2 is to maintain its relative weighting with respect to PC1, so that the resulting standard deviation for the OMI amplitude is less than 1.0 (Kiladis et al. 2014).

Another MJO index used in this study is the real-time multivariate MJO index (RMM) (Wheeler and Hendon 2004), which is available at http://www.bom.gov.au/climate/mjo/. The RMM is based on the zonal-mean zonal winds at 200 and 850 hPa and OLR over 15°S–15°N (Wheeler and Hendon 2004). Kiladis et al. (2014) showed that the RMM and OMI are highly correlated with each other, but the OMI is a better representative of the convective signal of the MJO. As PC1 and PC2 of the OMI are analogous to −PC2 and PC1 of the RMM, respectively (Kiladis et al. 2014), PC1 and PC2 of the OMI are expressed as and in this paper to help direct comparison with RMM. We conduct analyses using the RMM as well as the OMI, but the results mainly shown in this paper are done by using the OMI because our primary interest is on the convective component of the MJO. Comparisons of the results with the RMM and the OMI are discussed in section 4.

d. Other datasets

As a proxy for tropical convection, we use daily OLR data obtained from National Oceanic and Atmospheric Administration (NOAA) satellites with a 2.5°-grid spatial resolution (Gruber and Krueger 1984). The data period used in this study is January 1979–December 2013. We also use, as a proxy for tropical convection, the daily precipitation data estimated from the Tropical Rainfall Measuring Mission (TRMM) 3B43, version 7, satellite data (Huffman et al. 2007) from January 1998 to December 2013. The original data are provided with a 0.25° horizontal grid resolution, but in this study we reconstruct them with a 2.5° grid resolution, which is the same resolution as that of the OLR data. Other meteorological variables such as temperature, wind, divergence of horizontal wind, and specific humidity are from ERA-Interim data (Dee et al. 2011) from January 1979 to December 2013. We averaged the original data with a 6-h interval to daily time resolution.

3. Results

a. Time variations of ENSO, QBO, and MJO indices

Figure 2b shows a scatterplot of the QBO index (U50) versus the ENSO index (Niño-3.4 SSTA) for the months of austral summer (DJF); stars mark the months when the MJO events are identified, whereas diamonds mark the months without any MJO event. Here, an MJO event is defined as a period when the OMI amplitude is greater than or equal to one at phase 4 of the OMI, which is the phase when the active convective system is located over the eastern Indian Ocean through the Maritime Continent among eight phases (Wheeler and Hendon 2004). The composite results for these MJO events are shown in sections 3b and 3c. Gray lines divide the diagram into nine phases according to the QBO and ENSO phases. The number of months included in each phase is given in the figure, and the total number of months is 105. The number of months with MJO events is given in square brackets in each phase (the total number is 37). The correlation coefficient between the QBO and ENSO indices is 0.20 at zero lag. The correlation coefficients at other lags are smaller. This result is consistent with Garfinkel and Hartmann (2007) and Hu et al. (2012), who showed that there is no statistically significant relationship between the QBO and ENSO after the post-1979 satellite era. To separate the effect of ENSO from that of the QBO, the analyses in this paper are conducted only for DJF months in the neutral ENSO period. Using only the neutral ENSO period is also desirable to avoid the influence of ENSO on the OMI, which impacts a particular phase of the OMI during a strong event of either El Niño or La Niña. During the neutral ENSO period, there are 68 months, and during the E-QBO, N-QBO, and W-QBO phases, there are 23, 16, and 29 months, respectively.

The colored circles for the neutral ENSO condition in Fig. 1c imply that the OMI amplitude in DJF during the E-QBO phase (blue) is typically larger than those during the other QBO phases, and the amplitude during the W-QBO phase (red) is small with a few exceptions. Figure 3 shows the time variations of the daily during the neutral ENSO period, divided into the E-QBO (left), N-QBO (middle), and W-QBO (right) phases. The amplitude is generally large during austral summer and autumn (December–May), no matter what the QBO phase is. During DJF, the amplitude in the E-QBO phase is large, compared with other QBO phases, especially the W-QBO phase, with the notable exceptions of 1986 and 1988. During the other seasons, on the other hand, the amplitude in the W-QBO phase is comparable to that in the E-QBO phase. This is a simple plot which visualizes the seasonality of the MJO modulation by the QBO first reported by Yoo and Son (2016).

Fig. 3.
Fig. 3.

Time series of daily OMI amplitude during the neutral ENSO period, divided into the (left) E-QBO, (center) N-QBO, and (right) W-QBO phases. The amplitudes greater than or equal to one are filled with colors. Stars show the key days when the MJO events are identified.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

Figure 4 shows the first two OMI principal components, (abscissa) and (ordinate), plotted on the so-called Wheeler–Hendon phase diagrams (Wheeler and Hendon 2004) in each QBO phase, each ENSO phase, and all phases for DJF, and consecutive days are connected with a line. During the neutral ENSO period shown in the second row, the dots are scattered on the wide area of the plot in the E-QBO phase, and the number of days when the OMI amplitude is greater than or equal to 1.0 is about three-quarters of the total E-QBO days (523 days out of 691 days). In the W-QBO phase, on the other hand, the dots gather around the origin of the coordinate axes, and the days with account for about a half of the total W-QBO days (454 days out of 868 days). As a result, the mean OMI amplitude in each MJO phase is larger during the E-QBO phase than during the W-QBO phase, as shown by Yoo and Son (2016) in their Fig. 3d. This difference between the E-QBO and W-QBO phases is also discernible under El Niño or La Niña conditions, even though the numbers of days plotted on the phase diagrams are small.

Fig. 4.
Fig. 4.

Phase diagrams of (abscissa) and (ordinate) in the (left)–(right) E-QBO, N-QBO, W-QBO, and all QBO phases and (top)–(bottom) El Niño, neutral ENSO, La Niña, and all ENSO phase for DJF. The number of days included in each phase is indicated in each panel.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

b. Daily composite analysis of OLR

Composite analysis is performed with OMI phase diagram or amplitude , and phase of the OMI index. First, an MJO event is defined for a period with . Traditionally, time evolution of each MJO event is divided into eight phases: with for phase I, with . To make a composite of the MJO events in each phase I, the first day that satisfies both and is referred to as day 0 (i.e., the key day) with a separated period longer than 30 days from the previous key day and the next key day. The autocorrelation coefficient of the daily OMI amplitude falls to 0.3 at the 20-day lag, and it is 0.14 for the 30-day lag.

Figure 5 shows the composites of time evolution of the OMI for each phase of the MJO events during DJF in the neutral ENSO period. The composite is made for each QBO phase—that is, E-QBO (blue), N-QBO (green), and W-QBO (red)—for the 51-day period from 25 days before the key day (noted by a star) to 25 days after the key day. The total number of the identified MJO events N and those divided into each QBO phase (in parentheses) are noted in the figure. At whichever MJO phase, the OMI amplitude of the active events is typically larger in the E-QBO phase than in the W-QBO phase. The OMI amplitude in the N-QBO phase is the second largest in the cases when the key day is in phases 1–4. Duration of the MJO events with is also longer in the E-QBO phase in most of the MJO phases, except for phases 5 and 6.

Fig. 5.
Fig. 5.

Phase diagrams of the composite of the OMI for the E-QBO (blue), N-QBO (green), and W-QBO (red) phases of the MJO events during DJF in the neutral ENSO period. The key day is marked with a star, and the composite is made for the 51-day period of the key day plus and minus 25 days. The total number of the identified MJO events N and those divided into each QBO phase are noted in each panel.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

Phases 2–6 are referred to as the “wet phase” of MJO, during which the active convective system propagates eastward from the western Indian Ocean to the central Pacific. At phase 4, when the active convective system is located over the eastern Indian Ocean through the Maritime Continent (Wheeler and Hendon 2004), the total number of the identified MJO events is 23, the largest among the other phases of the wet phase of MJO, with 10, 6, and 7 in the E-QBO, N-QBO, and W-QBO phases, respectively. The key days are shown with stars in Fig. 3. In the rest of this subsection, we focus on the daily composite results for the MJO events identified in phase 4, using the unfiltered daily data. Regardless of the MJO phases in which the MJO events are identified, almost identical results are obtained even though differences between the E-QBO and W-QBO phases are smaller in magnitude.

The left panels in Figs. 6a–c show longitude–time sections of the composite OLR anomaly of the MJO events for the E-QBO, N-QBO, and W-QBO phases, respectively. The OLR values are averaged over 10°S–10°N and the anomaly from the climatological mean state during DJF in the neutral ENSO period (Fig. 6f) is plotted. The MJO convective center is defined for each day between days −25 and +25 as the location where the OLR anomaly has a minimum value less than −20 W m−2 (white crosses in Figs. 6a–c). Negative OLR anomalies propagate eastward in the Eastern Hemisphere, and they follow positive OLR anomalies before the key day and precede the positive ones after the key day. The composite amplitude of the anomalies is the largest in the E-QBO phase whereas it is the smallest in the W-QBO phase. The strong signal of convection lasts longer with a slower eastward-propagation speed in the E-QBO phase, compared with the other QBO phases [see also Son et al. (2017)]. Time series of the OMI phase and the phase number I (right panels in Figs. 6a–c) show that the MJO signal in the E-QBO phase takes about 50 days to circle around the phase diagram, whereas those in the N-QBO and W-QBO phases take about 41 and 44 days, respectively. In these QBO phases, the anomaly is small in the “dry phase” of MJO (I = 1, 7, and 8).

Fig. 6.
Fig. 6.

Longitude–time sections of the composite OLR anomaly (W m−2) of MJO events for the (a) E-QBO, (b) N-QBO, and (c) W-QBO phases, and for (d) the E–W composite difference of OLR, and of (e) the statistically significance of the composite difference. (f) The OLR values are averaged over 10°S–10°N and anomaly from the climatological mean state during DJF in the neutral ENSO period. The MJO convective center is marked with white crosses in (a)–(c). Time series of the OMI phase and the phase number I are shown to the right of (a)–(c).

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

Figure 6d shows the composite difference of OLR between the E-QBO and the W-QBO phases (herein, E–W composite difference), and Fig. 6e shows the statistical significance of the composite difference, which is determined by a two-sided Student’s t test at each longitude at each day. The composite OLR anomalies are significantly larger during the MJO active periods in the E-QBO phase than in the W-QBO phase from the eastern Indian Ocean to the central Pacific. The statistical significance of these differences is mostly around the 90% or 95% confidence level over the eastern Indian Ocean between days −15 and +5, whereas it is mostly over the 99% confidence level over the western and central Pacific after day +5.

Figures 7a–c and 7d–f show longitude–latitude sections of the composite OLR anomaly in the E-QBO and W-QBO phases, respectively, for three consecutive 10-day-averaged periods from days −15 to −5 (Fig. 7a and 7d), days −5 to +5 (Fig. 7b and 7e), and days +5 to +15 (Fig. 7c and 7f). The plotted data are the anomaly from the climatological mean state during DJF in the neutral ENSO period (Fig. 7j). In the climatological mean state, strong convective activity with low values of OLR is located in the Southern Hemisphere over Africa, from the eastern Indian Ocean to the western Pacific, and over South America. The eastward propagation of the negative OLR anomalies is confirmed again in Figs. 7a–c for the E-QBO phase and in Figs. 7d–f for the W-QBO phase. The negative OLR anomalies in the E-QBO phase are larger in magnitude and distributed over 20°S–20°N with a longitude width of about 50°. In the W-QBO phase, on the other hand, the negative anomalies are smaller in magnitude and distributed mainly in the Southern Hemisphere with the lesser meridional extent. The negative OLR anomalies in the W-QBO phase are located farther westward than those in the E-QBO phase for 10 days before the key day (Figs. 7a and 7d), and their locations become closer between both QBO phases in the following days (Figs. 7b and 7e). This implies the slower propagation speed in the E-QBO phase. Horizontal wind near the surface shows convergent pattern associated with the negative OLR anomalies in each panel.

Fig. 7.
Fig. 7.

Longitude–latitude sections of the composite OLR anomaly (W m−2; colors) and the composite anomalies of the horizontal winds at 900 hPa (m s−1; vectors) in the (a)–(c) E-QBO and (d)–(f) W-QBO phases, and (g)–(i) the E–W composite difference with statistical significance at a confidence level of 90% or greater, for the periods of (left) days −15 to −5, (center) days −5 to +5, and (right) days +5 to +15. The plotted data are anomalies from (j) the climatological mean state during DJF in the neutral ENSO period. The wind vector scales shown to the right of (j) are 5 m s−1 for the zonal and meridional winds.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

The E–W composite differences with statistical significance of 90%and higher are shown in Figs. 7g–i. Statistically significant E–W composite differences lie mostly between 10°S and 10°N, and they are discernible around the OLR anomalies related to the MJO convective activities—that is, negative differences over 90°–110°E from days −15 to −5, over 110°–130°E from days −5 to +5, and over 140°–170°E from days +5 to +15 with stronger magnitude. This indicates that the MJO convective activity is larger in the E-QBO phase than in the W-QBO phase. Between 160° and 80°W, significant positive difference is seen along 5°N, accompanied with negative difference in the north of it. These positive and negative E–W composite differences are due to southward displacement of the convectively active region over the intertropical convergence zone (ITCZ) in the W-QBO phase (Fig. 7d), compared with the E-QBO phase (Fig. 7a).

c. Daily composite analysis of other variables

Figure 8 shows longitude–time sections of the divergence of horizontal wind averaged for 300–100 hPa, the vertical wind for 500–250 hPa, the divergence of horizontal wind for 850–400 hPa, the specific humidity for 700–300 hPa, and the TRMM precipitation (from left to right) for the MJO events in the E-QBO (top row) and W-QBO (second row) phases, and the E–W composite difference with 90% statistical significance and more (third row). The data are averaged over 10°S–10°N and the anomalies from the climatological mean states during DJF in the neutral ENSO period (bottom row) are plotted. The composites are made for each QBO phase for the 51-day period centered on the key day in phase 4, by using the unfiltered daily data. The MJO convective centers defined in the previous subsection are marked with crosses in the figure. The anomalies related to the dynamical structure of the MJO exhibit the QBO modulation of eastward-propagation characteristics associated with the anomalies of convective activity related to the MJO: the divergence of horizontal wind in the upper troposphere (Figs. 8a and 8f) and the convergence in the lower troposphere (Figs. 8c and 8h), together with the upwelling in the middle troposphere (Figs. 8b and 8g), show good agreement in position and intensity with those of OLR anomaly (Figs. 6a and 6c). The anomalies are larger in the E-QBO phase than in the W-QBO phase, and these differences are statistically significant mostly above the 90% confidence level (Figs. 8k–m).

Fig. 8.
Fig. 8.

Longitude–time sections of the composite anomalies of (left)–(right) the divergence of horizontal wind averaged for 300–100 hPa (s−1), the vertical wind for 500–250 hPa (cm s−1), the divergence of horizontal wind for 850–400 hPa (s−1), the specific humidity for 700–300 hPa (kg kg−1), and the TRMM precipitation (mm day−1) for the MJO events in (a)–(e) the E-QBO and (f)–(j) W-QBO phases, and (k)–(o) the E–W composite difference with a statistical significance at the confidence level of 90% or greater. (p)–(t) The data are averaged over 10°S–10°N and anomalies from the climatological mean states during DJF in the neutral ENSO period. The MJO convective center is marked with crosses in (a)–(j). The color scale used for the vertical wind in (b),(g), and (l) is logarithmic.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

The anomalies related to the thermodynamic structure of the MJO also show the QBO modulation of eastward-propagation characteristics associated with the anomalies of convective activity related to the MJO. Positive and negative anomalies of specific humidity in the middle troposphere in the Eastern Hemisphere show good agreement in position and magnitude with those of the dynamical structure of the MJO as stated above, indicating vertical advection associated with the upwelling and downwelling of the MJO (Figs. 8d and 8i). Positive precipitation anomalies obtained by a rather independent TRMM dataset also show good agreement in position and magnitude with those of the dynamical structure related to the MJO (Figs. 8e and 8j). The anomalies of these variables are larger in the E-QBO phase than in the W-QBO phase, and the E–W differences are statistically significant mostly above the 90% confidence level (Figs. 8n and 8o).

The top and middle rows in Fig. 9 show longitude–height sections of the composites of the anomaly of specific humidity and the zonal and vertical winds in the E-QBO and W-QBO phases, respectively, for three consecutive 10-day-averaged periods from days −15 to −5, days −5 to +5, and days +5 to +15 (from left to right). The OLR anomalies are also shown at the bottom of each panel. The data are averaged over 10°S–10°N, and the anomalies from the climatological mean state during DJF in the neutral ENSO period (Fig. 9j) are shown. Positive anomalies of the specific humidity in the troposphere propagate eastward associated with the upwelling in the midtroposphere and horizontal convergence in the lower troposphere related to the MJO. Negative anomalies of the specific humidity associated with downward motions exist in front of the upwelling of the MJO convective center, and they are also seen in the rear of the MJO convective center from days +5 to +15. The magnitudes of the upwelling and the specific humidity anomaly around the MJO convective center are generally larger in the E-QBO phase than in the W-QBO phase.

Fig. 9.
Fig. 9.

Longitude–height sections of the composites anomalies of the specific humidity anomaly (kg kg−1; colors) and the zonal and vertical winds (m s−1 and cm s−1, respectively; vectors) in (a)–(c) the E-QBO and (d)–(f) W-QBO phases, and (g)–(i) the E–W composite difference with statistical significance at a confidence level of 90% or greater for the periods of (left) days −15 to −5, (left center) days −5 to +5, and (right center) days +5 to +15. The composite OLR anomaly (W m−2) is shown at the bottom of each panel. The data are averaged over 10°S–10°N and the anomaly from (j) the climatological mean state during DJF in the neutral ENSO period. In the E–W composite difference of the OLR, red marks show values exceeding statistical significance at the 90% confidence level. The color scales are logarithmic. The wind vector scales shown to the right of (c) are 20 m s−1 and 1 cm s−1 for the zonal and vertical winds, respectively.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

The E–W composite differences of the specific humidity in the longitude–height section are shown in the bottom row of Fig. 9 for the values with statistical significance with confidence levels of 90% and greater. In the E–W composite difference of the OLR in the bottom panel, statistically significant values larger than or equal to the 90% confidence level are marked by red circles. The E–W differences of the specific humidity and upward winds in the troposphere are statistically significant around the longitudes of the MJO convective center where the E–W composite difference of the OLR is significantly negative. This indicates that the upwelling and upward moisture transport around the MJO convective center are more enhanced in the E-QBO phase than in the W-QBO phase. Statistically significant E–W difference of the specific humidity with negative sign is also seen near the surface over 180°–80°W, where the ITCZ lies, from days −15 to −5, and the E–W difference is still discernible from days −5 to +5.

Figure 10 shows longitude–latitude sections of the composites of the temperature anomalies at 100 hPa in the E-QBO (top row) and W-QBO (second row) phases, and the E–W composite difference (third row) for three consecutive 10-day-averaged periods around the key day (from left to right). The temperature anomalies are from the climatological mean state during the DJF months in the neutral ENSO period (Fig. 10j). The climatological mean state shows nearly symmetric distribution about the equator with the lowest zonal-mean temperature on the equator even in austral summer. Low-temperature areas are confined over the western Pacific, forming the “horseshoe-shaped structure” (Newell and Gould-Stewart 1981; Highwood and Hoskins 1998; Fueglistaler et al. 2009; Nishimoto and Shiotani 2012, 2013). The temperature anomalies are also distributed nearly symmetric about the equator; low-temperature anomalies exist around the MJO convective center, over the Indian Ocean from days −15 to −5 (Figs. 10a and 10d), the western Pacific from days −5 to +5 (Figs. 10b and 10e), and the central Pacific from days +5 to +15 (Figs. 10c and 10f). The anomalies are larger in magnitude in wider area in the E-QBO phase than in the W-QBO phase. The E–W composite difference of the anomalies (Fig. 10g–i) is nearly symmetric about the ±10° equator in the tropics, with statistical significance of 90% and greater. The E–W composite differences are statistically significant in the east of the MJO convective center, showing lower temperatures in the E-QBO phase. From days +5 to +15, statistically significant E–W differences with negative sign are also seen in the tropics over 0°–110°E (Fig. 10i). In this area, positive anomalies with relatively large amplitude are located in the W-QBO phase (Fig. 10f). These temperature anomalies accompany descending anomalies at the same pressure level (not shown), suggesting the role of adiabatic heating process there.

Fig. 10.
Fig. 10.

As in Fig. 7, but for the temperature (K) at 100 hPa.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

d. Conditional sampling analysis

A composite analysis was performed in the previous subsections by focusing on the particular phase of the MJO (i.e., phase 4 of the OMI) after dividing into the E-QBO, N-QBO, and W-QBO phases, whereas in this subsection we perform a conditional sampling analysis focusing on the most active convective region for each day, irrespective of the MJO amplitude and phase, during DJF months in the neutral ENSO period. To detect and examine the center of an active convective region, we use the unfiltered daily data and search the minimum value of the OLR, , in the zonal and meridional directions between 10°S and 10°N to find its longitude and latitude . The signals treated in this subsection can involve mesoscale organization of convective systems or other convectively coupled waves besides the MJO, although the MJO signal may be dominant.

Figures 11a–c show histograms of relative frequency of , , and , respectively, for the E-QBO (blue) and W-QBO (red) phases with bin sizes of 5 W m−2, 15°, and 2.5°, respectively. The relative frequency (%) is given by dividing the number of samples in each bin by the total number (691 or 868 days for the E-QBO or W-QBO phases, respectively). The mean value of is 106.4 W m−2 for the E-QBO phase and 107.5 W m−2 for the W-QBO phase, and the difference is statistically significant with at the 90% confidence level. This result agrees with Garfinkel and Hartmann (2011) who used a general circulation model (GCM) and showed large differences in OLR between opposite QBO phases. The statistical significance is calculated by a two-sided Student’s t test with the effective sample size estimated by the number of independent samples (i.e., N′ = 432 and 540 for the E-QBO and W-QBO phases, respectively). Here, an independent sample is determined as the point that satisfies either 1) , 2) with , or 3) with and , where is the date of the nth independent sample. The longitudinal and latitudinal phase differences, and , are set to be 20° and 5°, respectively, where the correlation coefficient of the OLR with respect to the most active convective region falls to 0.3. The time difference is set to be 16 days because is constant with . The results are insensitive to the choice of this value. The frequency distributions of in the E-QBO and W-QBO phases show monomodal distributions with the peak at 105 and 110 W m−2, respectively. The frequency for the E-QBO phase is larger than that for the W-QBO phase for , and vice versa for . That is, the most active convection on each day is more active in the E-QBO phase than in the W-QBO phase.

Fig. 11.
Fig. 11.

Histograms of relative frequency (%) of (a) and its (b) and (c) for the E-QBO (blue) and W-QBO (red) phases with bin sizes of 5 W m−2, 15°, and 2.5°. The mean values for the E-QBO and W-QBO phases are denoted in each panel with red and blue dashed lines, respectively.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

Figure 11b shows that is mostly located over the Indian Ocean through the central Pacific (60°E–165°W), in addition to minor peaks over the African continent (~15°E) and over the South American continent (60°W), similar to the climatological distribution of the OLR value (Fig. 6f). The relative frequency distribution over the Indian Ocean through the central Pacific in the E-QBO phase shows a bimodal distribution with the primary peak at 165°E and another peak at 90°E, whereas that in the W-QBO phase shows a broad distribution with a peak at 165°E. Figure 11c shows that is mostly located in the Southern Hemisphere with a peak at 10°S. The frequency for the E-QBO phase is larger than that for the W-QBO phase in the Southern Hemisphere, whereas it is smaller in the Northern Hemisphere, except at 10°N.

To clarify characteristic features at the most active convection region, which can be different from those in relatively weak convection area, we examine dynamic and thermodynamic features of the conditional samples only at the most active convective region at for each day in the E-QBO and W-QBO phases and compare them with unconditional samples that are computed for all longitudinal grid points of the data over 10°S–10°N. Figures 12a–d show vertical profiles of the composite of the zonal wind u, the absolute value of the vertical shear of zonal wind, , the temperature, and the static stability, respectively, for the conditional samples for the E-QBO (blue) and W-QBO (red) phases in DJF in the neutral ENSO period. The unfiltered daily data are used without removing the climatological mean state, and the composite of is derived after the absolute value is calculated at each grid point. Whiskers in the figure denote the standard deviation. Figures 12e–h show the E–W composite difference for the conditional samples (black line with red circles) together with the E–W composite difference for the unconditional samples (gray line with blue circles). Statistical significance of the E–W composite difference for the conditional samples is calculated by a two-sided Student’s t test with the effective sample size N′ = 432 and 540 for the E-QBO and W-QBO phases, respectively, as mentioned before. For the unconditional samples, is assumed as the effective zonal and meridional grid numbers and , respectively, multiplied by the number of years (i.e., N′ = × × Nyr = 576 and 864 for the E-QBO and W-QBO phases, respectively). The longitudinal and latitudinal phase differences, and , are again set to be 20° and 5°, respectively, because the correlation coefficient of the OLR between any two points in the tropics falls to 0.3, when the phase differences become larger than those values. Statistical significance is marked with closed circles, open circles, and dots, which denote the 99%, 95%, and 90% confidence levels, respectively.

Fig. 12.
Fig. 12.

(a)–(d) Vertical profiles of the composites of the conditionally sampled data of zonal wind (m s−1), absolute value of zonal wind shear (s−1), temperature (K), and static stability (s−2), respectively, for the E-QBO (blue) and W-QBO (red) phases. Whiskers denote the ensemble mean plus and minus one standard deviation. (e)–(h) The E–W composite difference of the conditional samples (black line with red circles) and the unconditional samples (gray line with blue circles). The statistical significance of the E–W statistical differences is indicated with filled and open circles and dots denoting the 99%, 95%, and 90% confidence levels, respectively.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

There are noticeable differences in the composites of u and for the conditional samples of the E-QBO and W-QBO phases in the stratosphere (Figs. 12a and 12b). The E–W composite difference of u for the conditional samples (black line in Fig. 12e) is large and statistically significant in the stratosphere and is almost identical to that of the unconditional samples (gray line in Fig. 12e). On the other hand, the E–W composite difference of is largely different between the conditional and the unconditional samples in the lower stratosphere with statistical significance over the 99% confidence level at 70 hPa (Fig. 12f).

The composites of the temperature and the static stability for the conditional samples in the E-QBO and W-QBO phases are shown in Figs. 12c and 12d, respectively, and the E–W composite differences of these quantities are shown in Figs. 12g and 12h. The E–W composite difference of the temperature is significant and large (>1 K) in the stratosphere, being negative below 50 hPa with a peak at 70 hPa and positive at 30 hPa. The difference between the conditional and the unconditional samples is noticeable above 100 hPa, showing negatively larger values of the E–W composite difference for the conditional samples than for the unconditional samples. The corresponding significant E–W composite differences of the static stability are located above 175 hPa with negative and positive peaks at 100 and 50 hPa, respectively. Smaller static stability in the E-QBO phase than in the W-QBO phase is more prominent between 175 and 100 hPa for the conditional samples than for the unconditional samples. The lower temperature and the lower static stability in the upper troposphere are favorable conditions to develop intensive deep convection (Gray et al. 1992; Collimore et al. 2003; Nie and Sobel 2015). The E–W composite difference for the conditional samples shows statistical significance in some quantities near the surface, though the magnitudes are not so large except for the vertical shear of zonal wind.

Figures 13a–c show vertical profiles of the composites of the conditionally sampled data of vertical wind, divergence of horizontal wind, and specific humidity in the E-QBO (blue) and W-QBO (red) phases, and Figs. 13d–f show the E–W differences of these quantities. In the troposphere, although the statistically significance is not so high, the E–W composite differences of the vertical wind for the conditional samples show two peaks of diminished upwelling in the E-QBO phase than in the W-QBO phase, together with the corresponding negative and positive differences of the divergence above and below the negative peak of the difference of the vertical wind, respectively (Figs. 13d and 13e). These relationships between the upwelling and the divergence of the horizontal wind in the troposphere are dynamically consistent to hint modulation of active convection associated with the QBO phase. The E–W composite difference of the specific humidity shows statistically significant decrease near the surface in the E-QBO phase, compared with in the W-QBO phase (Fig. 13f). Wavy structures are seen in the E–W composite differences in the vertical wind above 100 hPa (Fig. 13d) and the divergence of horizontal wind above 150 hPa (Fig. 13e). The E–W differences are statistically significant in the vertical wind above 100 hPa, in the divergence of horizontal wind at 100 and 70 hPa, and in the specific humidity above 125 hPa (Fig. 13f). The E–W composite differences of these quantities for the unconditional samples (gray line) are small for the vertical wind and divergence of horizontal wind, and they are not statistically significant above the 90% confidence level. On the other hand, the E–W composite difference of the specific humidity for the unconditional samples (gray line) shows similar vertical profile for the conditional samples (black line) with comparable magnitudes, and they are also statistically significant in the lower stratosphere with negative signs.

Fig. 13.
Fig. 13.

As in Fig. 12, but for (a),(d) the vertical wind (cm s−1), (b),(e) the divergence (s−1), and (c),(f) the specific humidity (kg kg−1).

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

The top and second rows of Fig. 14 show longitude–height sections of the composite of the conditionally sampled data of zonal wind, vertical wind, and temperature anomaly (from left to right) centered at the most active convective region, , for the E-QBO and the W-QBO phases, respectively. The temperature anomaly is from the climatological zonal-mean state averaged over 10°S–10°N during DJF in the neutral ENSO period (see panel to the right of Fig. 14c). Wind vectors are overlaid in the tropospheric part and the scale is shown to the right of Fig. 14f. The OLR composite is also shown at the bottom of each panel. The zonal wind at 150 hPa in the upper troposphere has a negative peak at 15° west of and positive one at 90° east of (Figs. 14a and 14d). The corresponding peaks with the opposite sign exist below them near the surface. A group of upward motion around the most active convective region extends between −90° and +90° in the troposphere with a peak at (Figs. 14b and 14e). Positive temperature anomalies extend between −40 and +60° in the middle troposphere with a peak at (Figs. 14c and 14f). These dynamic and thermodynamic structures in the troposphere show similarity to the large-scale convective system associated with the MJO (Kiladis et al. 2005).

Fig. 14.
Fig. 14.

Longitude–height sections of the composite of the conditionally sampled data of (left) zonal wind (m s−1), (center) vertical wind (cm s−1), and (right) temperature anomaly (K) centered at the most active convective region for the (a)–(c) E-QBO and (d)–(f) W-QBO phases. (g)–(i) The E–W composite difference with statistical significance at the confidence level of 90% or greater. Vectors show the composites of horizontal and vertical wind in the troposphere (m s−1 and cm s−1, respectively). The temperature in (c) and (f) is the anomaly from (top right) the climatological zonal-mean state averaged over 10°S–10°N. The composite OLR (W m−2) is shown at the bottom of each panel with red circles denoting statistical significance at the 90% confidence level. Logarithmic color scales are used for the vertical wind in (a),(d), and (g). The reference vectors and color scale for the E–W composite differences in (g)–(i) are different than those for the composites in (a)–(f). The reference wind vectors for the composites shown to the right of (f) are 20 m s−1 and 3 cm s−1 for the horizontal and vertical winds, respectively, and those for the E–W composite difference shown to the right of (i) are 10 m s−1 and 0.5 cm s−1.

Citation: Journal of the Atmospheric Sciences 74, 4; 10.1175/JAS-D-16-0205.1

The bottom row of Fig. 14 shows the E–W composite differences with statistical significance at and above the 90% confidence level. Vectors of the E–W composite difference of the zonal and vertical wind components are overlaid in the troposphere if either one of the two components is statistically significant. The scale shown in the right of Fig. 14i is different from that used in the composite components (top and second rows). Statistically significant difference of the OLR with confidence level above 90% is marked with red circles. Statistically significant E–W composite differences of the zonal wind, vertical wind, and temperature anomaly over the 90% confidence level are rarely seen at the most active convective region in the lower to middle troposphere, as seen in Figs. 12 and 13. However, to the east of the most active convective region ( and ), the E–W differences of the OLR are statistically significant, mostly above the 95% confidence level. In this region, although they are not statistically significant above the 90% confidence level, except for the zonal wind, the E–W differences of the zonal wind, vertical wind, and temperature are positive in the middle troposphere (see Figs. 14g–i, and wind vectors for the vertical wind; the temperature difference is not seen because it is not statistically significant), together with positive and negative differences of the divergence of horizontal wind (not shown) in the upper and lower troposphere, respectively. These results suggest stronger deep convective systems in the E-QBO phase than in the W-QBO phase there. In addition, the E–W composite differences of the temperature and static stability (not shown) are statistically significant there around the tropopause, suggesting more favorable conditions to develop intensive deep convection in the E-QBO phase. On the other hand, to the west of the most active convective region over , the E–W composite differences of the zonal wind and vertical wind are positive in the middle troposphere, together with negative temperature difference (see Figs. 14g–i, and wind vectors for the vertical wind; the temperature difference is not shown because it is not statistically significant). This suggests the increase of adiabatic cooling process there in the E-QBO phase.

In both QBO phases, a wavy structure that tilts eastward with height from the upper troposphere to the lower stratosphere is noticeable in the vertical wind (Figs. 14b and 14e) and the temperature anomaly (Figs. 14c and 14f) around the most active convective region . The peak of the downward motion is located slightly west of at the 100-hPa level, whereas that of the negative temperature anomaly is located 10° east of at the same level. A similar wavy structure is also discernible in the easterly zonal wind with a peak located 10° west of at the 100-hPa level (Figs. 14a and 14d). According to these phase relationships, this wavy structure can be interpreted as a vertically propagating Kelvin response to convective heating (e.g., Andrews et al. 1987; Ryu et al. 2008). The E–W differences of these quantities are statistically significant around the most active convective region, with a negative sign in the zonal wind 10° west of at 100 hPa (Fig. 14g); positive and negative ones in the vertical wind with peaks to the west of at 100 hPa and to the east of at 70 hPa, respectively (Fig. 14h; see also Fig. 13d); and negatives one in the temperature at 100 hPa with a peak to the west of (Fig. 14i; see also Fig. 12g). These suggest that the wave amplitudes are significantly larger in the E-QBO phase than in the W-QBO phase. Above 70 hPa, the E–W composite differences of the zonal wind and temperature are statistically significant across all longitudes (Figs. 14g and 14i).

4. Discussion

We also conducted composite analyses using the RMM, instead of the OMI. The main difference from the OMI is that the RMM includes the large-scale circulation and convective features of the MJO, whereas the OMI only includes the large-scale convective features. The number of the MJO events identified in phase 4 of the RMM are 10, 3, and 8 for the E-QBO, N-QBO, and W-QBO phases, respectively, and about 90% of them are common to the MJO events identified by using the OMI. The composite OLR anomaly shows almost identical features to those used by the OMI (Fig. 6)—that is, larger negative value and slower eastward propagation with prolonged period of the active convection in the E-QBO phase than in the W-QBO phase. The E–W composite difference is also statistically significant from the eastern Indian Ocean to the central Pacific, but the statistically significant difference is more noticeable over the western Pacific around the key day, whereas it is less noticeable over the central Pacific between days +5 and +15. Yoo and Son (2016) showed in the supplementary material that daily correlation coefficients between the QBO and the RMM are smaller than those for the OMI, perhaps because bandpass filter is not applied to derive the RMM.

Significant correlation between the stratospheric QBO and the MJO was obtained only during austral summer by Yoo and Son (2016), and Fig. 3 is a simple plot to visualize the seasonality of the MJO modulation by the QBO. The MJO has seasonal variations in its strength and latitudinal locations (Zhang and Dong 2004; Zhang 2005) and in the propagation pattern (Kikuchi et al. 2012). In austral summer, the strongest MJO signals are located immediately south of the equator, showing the feature of eastward propagation. In boreal summer, on the other hand, the intraseasonal oscillation with northward propagation over the northern Indian Ocean is predominant, sometimes known as boreal summer intraseasonal oscillation (BSISO). As shown in Fig. 9 of Kikuchi et al. (2012), these two types of the propagation features appear in equinox seasons, and austral summer is only the season for the appearance of the eastward-propagating MJO. Further analyses are necessary to understand the seasonality of the relationship between the QBO and the MJO.

Tropical deep convection reaches around 300–100 hPa (Fueglistaler et al. 2009), and about 1% of tropical convective systems reach above 14 km (~150 hPa) while about 0.1% penetrate above 17 km (~100 hPa) (Liu and Zipser 2005). Such individual tropical deep convection can be affected by the dynamic or thermodynamic conditions above it. The present study showed that the convective activity at the most active convective region in each day is significantly stronger in the E-QBO phase than in the W-QBO phase (Fig. 11). The conditional sampling analysis focusing on the most active convective region showed that the E–W composite differences of the temperature and static stability are statistically significant around the tropopause and satisfy the favorable condition for the development of intensive deep convection around the tropopause more in the E-QBO phase than in the W-QBO phase (Figs. 12g and 12h). These features are also seen in the unconditional samples although the E–W composite difference is smaller. The QBO modulation of the tropical deep convection through the variation of the temperature and static stability around the tropical tropopause was shown by analyses of observation-based datasets (Huang et al. 2012) and a GCM simulation result (Garfinkel and Hartmann 2011). Nie and Sobel (2015) also demonstrated variations of the intensive deep convection in a cloud-system-resolving model simulation by imposing temperature variations around the tropical tropopause associated with the QBO on the climatological profile, as seen in Fig. 12g.

The abovementioned E–W composite differences of the temperature and static stability, which satisfy the favorable condition for the development of intensive deep convection, have larger amplitude over the most active convective region than other tropical regions, and the statistically significant differences extend into lower levels in the upper troposphere (cf. black and gray lines in Figs. 12g and 12h). At the same time, longitude–height sections of the vertical wind, zonal wind, and temperature centered at the most active convective region (Fig. 14) suggest that this amplification can be attributed to a vertically propagating Kelvin response (cf. Ryu et al. 2008). Figure 14 shows similar features to the large-scale convective system associated with the MJO (Kiladis et al. 2005). Compared with the longitude–height sections of the MJO whose convective center is located at 155°E [Figs. 3, 6, and 7 in Kiladis et al. (2005) for the zonal wind anomaly, mass flux, and temperature anomaly, respectively], there is no downwelling in the troposphere in the west of the most active convective region in Figs. 14b and 14e, whereas it is observed in Fig. 6 of Kiladis et al. (2005). The peak of temperature anomaly in the troposphere is located about 10° east of the convective center, similar to that in Fig. 7 of Kiladis et al. (2005), with much smaller magnitude (~0.6 K).

Although the conditional samples were collected irrespective of the MJO amplitude and phase, the composites of the conditional sampled data show an MJO-like structure. We estimate the number of samples that can be related to the MJO, using the longitude–latitude composite of OLR anomaly from the DJF climatology at each MJO phase. The percentage of samples that are located in the area with the low OLR anomaly (−5 W m−2) at the corresponding MJO phase is 52.5% and 44.2% for the E-QBO and the W-QBO phases, respectively, and at each MJO phase the percentage tends to be high in the wet phase of the MJO. For example, it is higher than 60% at phases 4–7 and higher than 50% at phases 3 and 8 for the E-QBO phase, whereas it is higher than 50% at phases 3–7 for the W-QBO phase. This suggests that about half of the signals treated in the conditional sampling analysis are related to the MJO, and the others can involve mesoscale organization of convective systems or other convectively coupled waves besides the MJO.

In Fig. 14, statistically significant E–W composite differences above the 90% confidence level are rarely seen at the most active convective region from the lower to middle troposphere, but they are discernible to the east of the most active convective region in the middle troposphere together with statistically significant E–W differences of the OLR. These results suggest the possible importance of multiscale interactions of moist convection through the downward influence of the stratospheric QBO on the troposphere. Here, composite analyses do not give any causality between the condition around the tropopause and the deep convective system associated with the MJO. Further studies are needed by both observational data analyses and numerical model experiments to clarify the causality, paying attention to multiscale interactions of moist convection, as the MJO consists of a hierarchy of convective activity with various horizontal and time scales (Nakazawa 1988; Zhang 2005).

5. Conclusions

The influence of the stratospheric QBO on the MJO during austral summer (DJF) and its statistical significance were investigated only for the period of neutral ENSO events, using NOAA OLR and ERA-Interim datasets for the period between January 1979 and December 2013, and TRMM precipitation dataset for the period between January 1998 and December 2013. The QBO phase was determined by the monthly mean zonal-mean zonal wind at 50 hPa (Fig. 1a), and the ENSO phase was determined by the sea surface temperature anomaly in the Niño-3.4 region (Fig. 1b). The number of DJF months in the neutral ENSO period is 68, among which the easterly QBO, neutral QBO, and westerly QBO phases account for 23, 16, and 29 months, respectively (Fig. 2b). The amplitude of the OLR-based MJO index (OMI) is generally large during December–May, no matter the QBO phase (Fig. 3). The OMI amplitude in DJF in the E-QBO phase is relatively large, compared with the other QBO phases, and that in the W-QBO phase is the smallest (Figs. 1c and 3). The amplitude difference among the QBO phases is not clear in the other seasons. The first two OMI principal components in DJF months were plotted on the Wheeler–Hendon phase diagrams divided into each QBO and ENSO phases (Fig. 4) and showed that in every MJO phase the OMI amplitude in the E-QBO phase is larger than that in the W-QBO phase. This is a simple visualization of the earlier finding of the stronger MJO signals in the E-QBO phase than in the W-QBO phase independent of the MJO phase (Yoo and Son 2016).

Daily composite analyses were performed to examine the MJO evolution in three QBO phases, by focusing on a particular phase of the MJO events, at phase 4 when the active convective systems are located over the eastern Indian Ocean through the Maritime Continent. The composite OLR anomaly in the E-QBO phase shows a large negative value, a slow eastward propagation, and a prolonged period of the active convection, compared with that in the W-QBO phase (Figs. 6 and 7). Statistically significant differences of the MJO activities between the E-QBO and W-QBO phases were also found in other dynamical variables related to the structure associated with the MJO, such as the divergence of horizontal wind and vertical wind, and in the moisture and the TRMM precipitation data with dynamical consistency (Figs. 8 and 9). The temperature at 100 hPa also shows significant E–W composite difference (Fig. 10).

A conditional sampling analysis was performed by focusing on the most active convective region for each day, irrespective of the MJO amplitude and phase. As the center of the active convection, the location of minimum OLR was detected for each day in the tropics (0°–360°) between 10°S and 10°N by using the unfiltered daily data. Histograms of relative frequency of the minimum OLR value and its location in longitude and latitude show noticeable differences between the E-QBO and W-QBO phases (Fig. 11). Vertical profiles of the composites of the conditionally sampled data over the most active convective region show lower temperature and static stability around the tropopause in the E-QBO phase, compared with those in the W-QBO phase, and this feature is more prominent and extends into lower levels in the upper troposphere in the conditional samples than in the unconditional samples (Figs. 12 and 13). These E–W composite differences indicate that the thermodynamic condition around the tropopause above the most active convective region is more favorable to develop deep convection in the E-QBO phase than in the W-QBO phase, though composite analyses do not give any causality between the condition around the tropopause and the deep convective system.

The composite longitude–height sections of the zonal and vertical winds and temperature anomaly relative to the location of minimum OLR suggest that the amplification of the preferable thermodynamic condition above the active convective region can be attributed to a vertically propagating Kelvin response to the convective heating (Fig. 14). In addition, Fig. 14 shows similar features of the large-scale convective system associated with the MJO in each QBO phase, and the magnitudes of these variables are larger in the E-QBO phase than in the W-QBO phase, even though the conditional samples were collected irrespective of the MJO amplitude and phase. The percentage of samples that can be related to the MJO is 52.5% and 44.2% for the E-QBO and W-QBO phases, respectively, as described in the discussion section. This suggests that about half of the signals treated in the conditional sampling analysis are related to the MJO, and the others can involve mesoscale organization of convective systems or other convectively coupled waves besides the MJO. Statistically significant E–W composite differences of the conditionally sampled data are rarely seen in the lower to middle troposphere at the most active convective region, but they are discernible on the east of the most active convective region in the middle troposphere together with statistically significant E–W differences of the OLR. This suggests the possible importance of multiscale interactions of moist convection through the downward influence of the stratospheric QBO on the troposphere.

Acknowledgments

The authors thank Marv Geller, Seok-Woo Son, Harry Hendon, George Kiladis, and two anonymous reviewers for their helpful comments and suggestions. This work was supported by JSPS KAKENHI (S) Grant 24224011 and JSPS Core-to-Core Program, B. Asia-Africa Science Platforms.

REFERENCES

  • Albers, J. R., G. N. Kiladis, T. Birner, and J. Dias, 2016: Tropical upper-tropospheric potential vorticity intrusions during sudden stratospheric warmings. J. Atmos. Sci., 73, 23612384, doi:10.1175/JAS-D-15-0238.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andrews, D. G., J. R. Holton, and C. B. Leovy, 1987: Middle Atmosphere Dynamics. International Geophysics Series, Vol. 40, Academic Press, 489 pp.

  • Baldwin, M. P., and T. J. Dunkerton, 2001: Stratospheric harbingers of anomalous weather regimes. Science, 294, 581584, doi:10.1126/science.1063315.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Baldwin, M. P., and Coauthors, 2001: The quasi-biennial oscillation. Rev. Geophys., 39, 179229, doi:10.1029/1999RG000073.

  • Collimore, C. C., M. H. Hitchman, and D. W. Martin, 1998: Is there a quasi-biennial oscillation in tropical deep convection? Geophys. Res. Lett., 25, 333336, doi:10.1029/97GL03722.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collimore, C. C., D. W. Martin, M. H. Hitchman, A. Huesmann, and D. E. Waliser, 2003: On the relationship between the QBO and tropical deep convection. J. Climate, 16, 25522568, doi:10.1175/1520-0442(2003)016<2552:OTRBTQ>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eguchi, N., and K. Kodera, 2010: Impacts of stratospheric sudden warming event on tropical clouds and moisture fields in the TTL: A case study. SOLA, 6, 137140, doi:10.2151/sola.2010-035.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Emanuel, K., S. Solomon, D. Folini, S. Davis, and C. Cagnazzo, 2013: Influence of tropical tropopause layer cooling on Atlantic hurricane activity. J. Climate, 26, 22882301, doi:10.1175/JCLI-D-12-00242.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fueglistaler, S., A. E. Dessler, T. J. Dunkerton, I. Folkins, Q. Fu, and P. W. Mote, 2009: Tropical tropopause layer. Rev. Geophys., 47, RG1004, doi:10.1029/2008RG000267.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Garfinkel, C. I., and D. L. Hartmann, 2007: Effects of the El Niño–Southern Oscillation and the quasi-biennial oscillation on polar temperatures in the stratosphere. J. Geophys. Res., 112, D19112, doi:10.1029/2007JD008481.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Garfinkel, C. I., and D. L. Hartmann, 2011: The influence of the quasi-biennial oscillation on the troposphere in winter in a hierarchy of models. Part II: Perpetual winter WACCM runs. J. Atmos. Sci., 68, 20262041, doi:10.1175/2011JAS3702.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giorgetta, M. A., L. Bengtsson, and K. Arpe, 1999: An investigation of QBO signals in the east Asian and Indian monsoon in GCM experiments. Climate Dyn., 15, 435450, doi:10.1007/s003820050292.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gray, W. M., J. D. Sheaffer, and J. A. Knaff, 1992: Hypothesized mechanism for stratospheric QBO influence on ENSO variability. Geophys. Res. Lett., 19, 107110, doi:10.1029/91GL02950.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gruber, A., and A. F. Krueger, 1984: The status of the NOAA outgoing longwave radiation data set. Bull. Amer. Meteor. Soc., 65, 958962, doi:10.1175/1520-0477(1984)065<0958:TSOTNO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Highwood, E. J., and B. J. Hoskins, 1998: The tropical tropopause. Quart. J. Roy. Meteor. Soc., 124, 15791604, doi:10.1002/qj.49712454911.

  • Hu, Z.-Z. Z., B. Huang, J. L. Kinter, Z. Wu, and A. Kumar, 2012: Connection of the stratospheric QBO with global atmospheric general circulation and tropical SST. Part II: Interdecadal variations. Climate Dyn., 38, 2543, doi:10.1007/s00382-011-1073-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, B., Z.-Z. Hu, J. L. Kinter III, Z. Wu, and A. Kumar, 2012: Connection of stratospheric QBO with global atmospheric general circulation and tropical SST. Part I: Methodology and composite life cycle. Climate Dyn., 38, 123, doi:10.1007/s00382-011-1250-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and Coauthors, 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 3855, doi:10.1175/JHM560.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kikuchi, K., B. Wang, and Y. Kajikawa, 2012: Bimodal representation of the tropical intraseasonal oscillation. Climate Dyn., 38, 19892000, doi:10.1007/s00382-011-1159-1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kiladis, G. N., K. H. Straub, and P. T. Haertel, 2005: Zonal and vertical structure of the Madden–Julian oscillation. J. Atmos. Sci., 62, 27902809, doi:10.1175/JAS3520.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 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, 16971715, doi:10.1175/MWR-D-13-00301.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kodera, K., H. Mukougawa, and Y. Kuroda, 2011: A general circulation model study of the impact of a stratospheric sudden warming event on tropical convection. SOLA, 7, 197200, doi:10.2151/sola.2011-050.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kuma, K.-I., 1990: A quasi-biennial oscillation in the intensity of the intra-seasonal oscillation. Int. J. Climatol., 10, 263278, doi:10.1002/joc.3370100304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liess, S., and M. A. Geller, 2012: On the relationship between QBO and distribution of tropical deep convection. J. Geophys. Res., 117, D03108, doi:10.1029/2011JD016317.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., and E. J. Zipser, 2005: Global distribution of convection penetrating the tropical tropopause. J. Geophys. Res., 110, D23104, doi:10.1029/2005JD006063.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., B. Tian, K.-F. Li, G. L. Manney, N. J. Livesey, Y. L. Yung, and D. E. Waliser, 2014: Northern Hemisphere mid-winter vortex-displacement and vortex-split stratospheric sudden warmings: Influence of the Madden-Julian Oscillation and Quasi-Biennial Oscillation. J. Geophys. Res. Atmos., 119, 12 59912 620, doi:10.1002/2014JD021876.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nakazawa, T., 1988: Tropical super clusters within intraseasonal variations over the western Pacific. J. Meteor. Soc. Japan, 66, 823839.

  • Newell, R. E., and S. Gould-Stewart, 1981: A stratospheric fountain? J. Atmos. Sci., 38, 27892796, doi:10.1175/1520-0469(1981)038<2789:ASF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nie, J., and A. H. Sobel, 2015: Responses of tropical deep convection to the QBO: Cloud-resolving simulations. J. Atmos. Sci., 72, 36253638, doi:10.1175/JAS-D-15-0035.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nishimoto, E., and M. Shiotani, 2012: Seasonal and interannual variability in the temperature structure around the tropical tropopause and its relationship with convective activities. J. Geophys. Res., 117, D02104, doi:10.1029/2011JD016936.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nishimoto, E., and M. Shiotani, 2013: Intraseasonal variations in the tropical tropopause temperature revealed by cluster analysis of convective activity. J. Geophys. Res. Atmos., 118, 35453556, doi:10.1002/jgrd.50281.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nishimoto, E., S. Yoden, and H.-H. Bui, 2016: Vertical momentum transports associated with moist convection and gravity waves in a minimal model of QBO-like oscillation. J. Atmos. Sci., 73, 29352957, doi:10.1175/JAS-D-15-0265.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Plumb, A. R., and R. C. Bell, 1982: A model of the quasi-biennial oscillation on an equatorial beta-plane. Quart. J. Roy. Meteor. Soc., 108, 335352, doi:10.1002/qj.49710845604.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryu, J.-H., S. Lee, and S.-W. Son, 2008: Vertically propagating Kelvin waves and tropical tropopause variability. J. Atmos. Sci., 65, 18171837, doi:10.1175/2007JAS2466.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Son, S.-W., Y. Lim, C. Yoo, H. H. Hendon, and J. Kim, 2017: Stratospheric control of Madden–Julian oscillation. J. Climate, 30, 19091922, doi:10.1175/JCLI-D-16-0620.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wheeler, M. C., 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, 19171932, doi:10.1175/1520-0493(2004)132<1917:AARMMI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yoden, S., M. Taguchi, and Y. Naito, 2002: Numerical studies on time variations of the troposphere-stratosphere coupled system. J. Meteor. Soc. Japan, 80, 811830, doi:10.2151/jmsj.80.811.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yoden, S., H.-H. Bui, and E. Nishimoto, 2014: A minimal model of QBO-like oscillation in a stratosphere-troposphere coupled system under a radiative-moist convective quasi-equilibrium state. SOLA, 10, 112116, doi:10.2151/sola.2014-023.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yoo, C., and S.-W. Son, 2016: Modulation of the boreal wintertime Madden–Julian oscillation by the stratospheric quasi-biennial oscillation. Geophys. Res. Lett., 43, 13921398, doi:10.1002/2016GL067762.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, C., 2005: Madden-Julian Oscillation. Rev. Geophys., 43, RG2003, doi:10.1029/2004RG000158.

  • Zhang, C., and M. Dong, 2004: Seasonality in the Madden–Julian oscillation. J. Climate, 17, 31693180, doi:10.1175/1520-0442(2004)017<3169:SITMO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save
  • Albers, J. R., G. N. Kiladis, T. Birner, and J. Dias, 2016: Tropical upper-tropospheric potential vorticity intrusions during sudden stratospheric warmings. J. Atmos. Sci., 73, 23612384, doi:10.1175/JAS-D-15-0238.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andrews, D. G., J. R. Holton, and C. B. Leovy, 1987: Middle Atmosphere Dynamics. International Geophysics Series, Vol. 40, Academic Press, 489 pp.

  • Baldwin, M. P., and T. J. Dunkerton, 2001: Stratospheric harbingers of anomalous weather regimes. Science, 294, 581584, doi:10.1126/science.1063315.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Baldwin, M. P., and Coauthors, 2001: The quasi-biennial oscillation. Rev. Geophys., 39, 179229, doi:10.1029/1999RG000073.

  • Collimore, C. C., M. H. Hitchman, and D. W. Martin, 1998: Is there a quasi-biennial oscillation in tropical deep convection? Geophys. Res. Lett., 25, 333336, doi:10.1029/97GL03722.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collimore, C. C., D. W. Martin, M. H. Hitchman, A. Huesmann, and D. E. Waliser, 2003: On the relationship between the QBO and tropical deep convection. J. Climate, 16, 25522568, doi:10.1175/1520-0442(2003)016<2552:OTRBTQ>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eguchi, N., and K. Kodera, 2010: Impacts of stratospheric sudden warming event on tropical clouds and moisture fields in the TTL: A case study. SOLA, 6, 137140, doi:10.2151/sola.2010-035.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Emanuel, K., S. Solomon, D. Folini, S. Davis, and C. Cagnazzo, 2013: Influence of tropical tropopause layer cooling on Atlantic hurricane activity. J. Climate, 26, 22882301, doi:10.1175/JCLI-D-12-00242.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fueglistaler, S., A. E. Dessler, T. J. Dunkerton, I. Folkins, Q. Fu, and P. W. Mote, 2009: Tropical tropopause layer. Rev. Geophys., 47, RG1004, doi:10.1029/2008RG000267.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Garfinkel, C. I., and D. L. Hartmann, 2007: Effects of the El Niño–Southern Oscillation and the quasi-biennial oscillation on polar temperatures in the stratosphere. J. Geophys. Res., 112, D19112, doi:10.1029/2007JD008481.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Garfinkel, C. I., and D. L. Hartmann, 2011: The influence of the quasi-biennial oscillation on the troposphere in winter in a hierarchy of models. Part II: Perpetual winter WACCM runs. J. Atmos. Sci., 68, 20262041, doi:10.1175/2011JAS3702.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giorgetta, M. A., L. Bengtsson, and K. Arpe, 1999: An investigation of QBO signals in the east Asian and Indian monsoon in GCM experiments. Climate Dyn., 15, 435450, doi:10.1007/s003820050292.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gray, W. M., J. D. Sheaffer, and J. A. Knaff, 1992: Hypothesized mechanism for stratospheric QBO influence on ENSO variability. Geophys. Res. Lett., 19, 107110, doi:10.1029/91GL02950.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gruber, A., and A. F. Krueger, 1984: The status of the NOAA outgoing longwave radiation data set. Bull. Amer. Meteor. Soc., 65, 958962, doi:10.1175/1520-0477(1984)065<0958:TSOTNO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Highwood, E. J., and B. J. Hoskins, 1998: The tropical tropopause. Quart. J. Roy. Meteor. Soc., 124, 15791604, doi:10.1002/qj.49712454911.

  • Hu, Z.-Z. Z., B. Huang, J. L. Kinter, Z. Wu, and A. Kumar, 2012: Connection of the stratospheric QBO with global atmospheric general circulation and tropical SST. Part II: Interdecadal variations. Climate Dyn., 38, 2543, doi:10.1007/s00382-011-1073-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, B., Z.-Z. Hu, J. L. Kinter III, Z. Wu, and A. Kumar, 2012: Connection of stratospheric QBO with global atmospheric general circulation and tropical SST. Part I: Methodology and composite life cycle. Climate Dyn., 38, 123, doi:10.1007/s00382-011-1250-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and Coauthors, 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 3855, doi:10.1175/JHM560.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kikuchi, K., B. Wang, and Y. Kajikawa, 2012: Bimodal representation of the tropical intraseasonal oscillation. Climate Dyn., 38, 19892000, doi:10.1007/s00382-011-1159-1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kiladis, G. N., K. H. Straub, and P. T. Haertel, 2005: Zonal and vertical structure of the Madden–Julian oscillation. J. Atmos. Sci., 62, 27902809, doi:10.1175/JAS3520.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 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, 16971715, doi:10.1175/MWR-D-13-00301.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kodera, K., H. Mukougawa, and Y. Kuroda, 2011: A general circulation model study of the impact of a stratospheric sudden warming event on tropical convection. SOLA, 7, 197200, doi:10.2151/sola.2011-050.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kuma, K.-I., 1990: A quasi-biennial oscillation in the intensity of the intra-seasonal oscillation. Int. J. Climatol., 10, 263278, doi:10.1002/joc.3370100304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liess, S., and M. A. Geller, 2012: On the relationship between QBO and distribution of tropical deep convection. J. Geophys. Res., 117, D03108, doi:10.1029/2011JD016317.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., and E. J. Zipser, 2005: Global distribution of convection penetrating the tropical tropopause. J. Geophys. Res., 110, D23104, doi:10.1029/2005JD006063.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., B. Tian, K.-F. Li, G. L. Manney, N. J. Livesey, Y. L. Yung, and D. E. Waliser, 2014: Northern Hemisphere mid-winter vortex-displacement and vortex-split stratospheric sudden warmings: Influence of the Madden-Julian Oscillation and Quasi-Biennial Oscillation. J. Geophys. Res. Atmos., 119, 12 59912 620, doi:10.1002/2014JD021876.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nakazawa, T., 1988: Tropical super clusters within intraseasonal variations over the western Pacific. J. Meteor. Soc. Japan, 66, 823839.

  • Newell, R. E., and S. Gould-Stewart, 1981: A stratospheric fountain? J. Atmos. Sci., 38, 27892796, doi:10.1175/1520-0469(1981)038<2789:ASF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nie, J., and A. H. Sobel, 2015: Responses of tropical deep convection to the QBO: Cloud-resolving simulations. J. Atmos. Sci., 72, 36253638, doi:10.1175/JAS-D-15-0035.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nishimoto, E., and M. Shiotani, 2012: Seasonal and interannual variability in the temperature structure around the tropical tropopause and its relationship with convective activities. J. Geophys. Res., 117, D02104, doi:10.1029/2011JD016936.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nishimoto, E., and M. Shiotani, 2013: Intraseasonal variations in the tropical tropopause temperature revealed by cluster analysis of convective activity. J. Geophys. Res. Atmos., 118, 35453556, doi:10.1002/jgrd.50281.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nishimoto, E., S. Yoden, and H.-H. Bui, 2016: Vertical momentum transports associated with moist convection and gravity waves in a minimal model of QBO-like oscillation. J. Atmos. Sci., 73, 29352957, doi:10.1175/JAS-D-15-0265.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Plumb, A. R., and R. C. Bell, 1982: A model of the quasi-biennial oscillation on an equatorial beta-plane. Quart. J. Roy. Meteor. Soc., 108, 335352, doi:10.1002/qj.49710845604.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryu, J.-H., S. Lee, and S.-W. Son, 2008: Vertically propagating Kelvin waves and tropical tropopause variability. J. Atmos. Sci., 65, 18171837, doi:10.1175/2007JAS2466.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Son, S.-W., Y. Lim, C. Yoo, H. H. Hendon, and J. Kim, 2017: Stratospheric control of Madden–Julian oscillation. J. Climate, 30, 19091922, doi:10.1175/JCLI-D-16-0620.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wheeler, M. C., 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, 19171932, doi:10.1175/1520-0493(2004)132<1917:AARMMI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yoden, S., M. Taguchi, and Y. Naito, 2002: Numerical studies on time variations of the troposphere-stratosphere coupled system. J. Meteor. Soc. Japan, 80, 811830, doi:10.2151/jmsj.80.811.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yoden, S., H.-H. Bui, and E. Nishimoto, 2014: A minimal model of QBO-like oscillation in a stratosphere-troposphere coupled system under a radiative-moist convective quasi-equilibrium state. SOLA, 10, 112116, doi:10.2151/sola.2014-023.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yoo, C., and S.-W. Son, 2016: Modulation of the boreal wintertime Madden–Julian oscillation by the stratospheric quasi-biennial oscillation. Geophys. Res. Lett., 43, 13921398, doi:10.1002/2016GL067762.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, C., 2005: Madden-Julian Oscillation. Rev. Geophys., 43, RG2003, doi:10.1029/2004RG000158.

  • Zhang, C., and M. Dong, 2004: Seasonality in the Madden–Julian oscillation. J. Climate, 17, 31693180, doi:10.1175/1520-0442(2004)017<3169:SITMO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    Time series of (a) U50, (b) Niño-3.4 SSTA, and (c) OMI amplitude for January 1979–December 2013. Dashed lines in (a) and (b) show the thresholds separating the QBO phases and ENSO phases, respectively. DJF months are shaded in gray. DJF months in the E-QBO, N-QBO, and W-QBO phases are marked by blue, green, and red color symbols, respectively, and those in the El Niño, neutral ENSO, and La Niña periods are marked with crosses, circles, and triangles, respectively.

  • Fig. 2.

    Histograms of (a) U50 and (c) Niño-3.4 SSTA and (b) scatterplot of U50 vs Niño-3.4 SSTA for DJF months. Stars mark the months when the MJO events are identified, whereas diamonds mark the months without any MJO event. Gray lines show the thresholds separating ENSO or QBO phases. At each phase, the number of months included in each phase is indicated, and the number of months with MJO events is given in square brackets.

  • Fig. 3.

    Time series of daily OMI amplitude during the neutral ENSO period, divided into the (left) E-QBO, (center) N-QBO, and (right) W-QBO phases. The amplitudes greater than or equal to one are filled with colors. Stars show the key days when the MJO events are identified.

  • Fig. 4.

    Phase diagrams of (abscissa) and (ordinate) in the (left)–(right) E-QBO, N-QBO, W-QBO, and all QBO phases and (top)–(bottom) El Niño, neutral ENSO, La Niña, and all ENSO phase for DJF. The number of days included in each phase is indicated in each panel.

  • Fig. 5.

    Phase diagrams of the composite of the OMI for the E-QBO (blue), N-QBO (green), and W-QBO (red) phases of the MJO events during DJF in the neutral ENSO period. The key day is marked with a star, and the composite is made for the 51-day period of the key day plus and minus 25 days. The total number of the identified MJO events N and those divided into each QBO phase are noted in each panel.

  • Fig. 6.

    Longitude–time sections of the composite OLR anomaly (W m−2) of MJO events for the (a) E-QBO, (b) N-QBO, and (c) W-QBO phases, and for (d) the E–W composite difference of OLR, and of (e) the statistically significance of the composite difference. (f) The OLR values are averaged over 10°S–10°N and anomaly from the climatological mean state during DJF in the neutral ENSO period. The MJO convective center is marked with white crosses in (a)–(c). Time series of the OMI phase and the phase number I are shown to the right of (a)–(c).

  • Fig. 7.

    Longitude–latitude sections of the composite OLR anomaly (W m−2; colors) and the composite anomalies of the horizontal winds at 900 hPa (m s−1; vectors) in the (a)–(c) E-QBO and (d)–(f) W-QBO phases, and (g)–(i) the E–W composite difference with statistical significance at a confidence level of 90% or greater, for the periods of (left) days −15 to −5, (center) days −5 to +5, and (right) days +5 to +15. The plotted data are anomalies from (j) the climatological mean state during DJF in the neutral ENSO period. The wind vector scales shown to the right of (j) are 5 m s−1 for the zonal and meridional winds.

  • Fig. 8.

    Longitude–time sections of the composite anomalies of (left)–(right) the divergence of horizontal wind averaged for 300–100 hPa (s−1), the vertical wind for 500–250 hPa (cm s−1), the divergence of horizontal wind for 850–400 hPa (s−1), the specific humidity for 700–300 hPa (kg kg−1), and the TRMM precipitation (mm day−1) for the MJO events in (a)–(e) the E-QBO and (f)–(j) W-QBO phases, and (k)–(o) the E–W composite difference with a statistical significance at the confidence level of 90% or greater. (p)–(t) The data are averaged over 10°S–10°N and anomalies from the climatological mean states during DJF in the neutral ENSO period. The MJO convective center is marked with crosses in (a)–(j). The color scale used for the vertical wind in (b),(g), and (l) is logarithmic.

  • Fig. 9.

    Longitude–height sections of the composites anomalies of the specific humidity anomaly (kg kg−1; colors) and the zonal and vertical winds (m s−1 and cm s−1, respectively; vectors) in (a)–(c) the E-QBO and (d)–(f) W-QBO phases, and (g)–(i) the E–W composite difference with statistical significance at a confidence level of 90% or greater for the periods of (left) days −15 to −5, (left center) days −5 to +5, and (right center) days +5 to +15. The composite OLR anomaly (W m−2) is shown at the bottom of each panel. The data are averaged over 10°S–10°N and the anomaly from (j) the climatological mean state during DJF in the neutral ENSO period. In the E–W composite difference of the OLR, red marks show values exceeding statistical significance at the 90% confidence level. The color scales are logarithmic. The wind vector scales shown to the right of (c) are 20 m s−1 and 1 cm s−1 for the zonal and vertical winds, respectively.

  • Fig. 10.

    As in Fig. 7, but for the temperature (K) at 100 hPa.

  • Fig. 11.

    Histograms of relative frequency (%) of (a) and its (b) and (c) for the E-QBO (blue) and W-QBO (red) phases with bin sizes of 5 W m−2, 15°, and 2.5°. The mean values for the E-QBO and W-QBO phases are denoted in each panel with red and blue dashed lines, respectively.

  • Fig. 12.

    (a)–(d) Vertical profiles of the composites of the conditionally sampled data of zonal wind (m s−1), absolute value of zonal wind shear (s−1), temperature (K), and static stability (s−2), respectively, for the E-QBO (blue) and W-QBO (red) phases. Whiskers denote the ensemble mean plus and minus one standard deviation. (e)–(h) The E–W composite difference of the conditional samples (black line with red circles) and the unconditional samples (gray line with blue circles). The statistical significance of the E–W statistical differences is indicated with filled and open circles and dots denoting the 99%, 95%, and 90% confidence levels, respectively.

  • Fig. 13.

    As in Fig. 12, but for (a),(d) the vertical wind (cm s−1), (b),(e) the divergence (s−1), and (c),(f) the specific humidity (kg kg−1).

  • Fig. 14.

    Longitude–height sections of the composite of the conditionally sampled data of (left) zonal wind (m s−1), (center) vertical wind (cm s−1), and (right) temperature anomaly (K) centered at the most active convective region for the (a)–(c) E-QBO and (d)–(f) W-QBO phases. (g)–(i) The E–W composite difference with statistical significance at the confidence level of 90% or greater. Vectors show the composites of horizontal and vertical wind in the troposphere (m s−1 and cm s−1, respectively). The temperature in (c) and (f) is the anomaly from (top right) the climatological zonal-mean state averaged over 10°S–10°N. The composite OLR (W m−2) is shown at the bottom of each panel with red circles denoting statistical significance at the 90% confidence level. Logarithmic color scales are used for the vertical wind in (a),(d), and (g). The reference vectors and color scale for the E–W composite differences in (g)–(i) are different than those for the composites in (a)–(f). The reference wind vectors for the composites shown to the right of (f) are 20 m s−1 and 3 cm s−1 for the horizontal and vertical winds, respectively, and those for the E–W composite difference shown to the right of (i) are 10 m s−1 and 0.5 cm s−1.