Modulation of Madden-Julian Oscillation Activity by the Tropical Pacific-Indian Ocean Associated Mode

Lifeng LI, Xin LI, Xiong CHEN*, Chongyin LI,2, Jianqi ZHANG, and Yulong SHAN

1College of Meteorology and Oceanography, National University of Defense Technology, Nanjing 211101, China

2State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics,Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China

(Received 5 January 2020; revised 18 August 2020; accepted 21 August 2020)

ABSTRACT In this study, the impacts of the tropical Pacific—Indian Ocean associated mode (PIOAM) on Madden—Julian Oscillation (MJO) activity were investigated using reanalysis data. In the positive (negative) phase of the PIOAM, the amplitudes of MJO zonal wind and outgoing longwave radiation are significantly weakened (enhanced) over the Indian Ocean, while they are enhanced (weakened) over the central and eastern Pacific. The eastward propagation of the MJO can extend to the central Pacific in the positive phase of the PIOAM, whereas it is mainly confined to west of 160°E in the negative phase. The PIOAM impacts MJO activity by modifying the atmospheric circulation and moisture budget.Anomalous ascending (descending) motion and positive (negative) moisture anomalies occur over the western Indian Ocean and central-eastern Pacific (Maritime Continent and western Pacific) during the positive phase of the PIOAM. The anomalous circulation is almost the opposite in the negative phases of the PIOAM. This anomalous circulation and moisture can modulate the activity of the MJO. The stronger moistening over the Indian Ocean induced by zonal and vertical moisture advection leads to the stronger MJO activity over the Indian Ocean in the negative phase of the PIOAM.During the positive phase of the PIOAM, the MJO propagates farther east over the central Pacific owing to the stronger moistening there, which is mainly attributable to the meridional and vertical moisture advection, especially low-frequency background state moisture advection by the MJO’s meridional and vertical velocities.

Key words: tropical Pacific—Indian Ocean associated mode, MJO, moisture budget analysis, anomalous circulation

1. Introduction

The Madden—Julian Oscillation (MJO; Madden and Julian, 1971, 1972) is the dominant component of tropical atmospheric intraseasonal variability. It is a planetary-scale circulation anomaly coupled with convection, and it exhibits significant interannual and interdecadal variations (Madden and Julian, 1994; Chen et al., 2017). As the bridge connecting weather and climate variations, the MJO has significant impacts on the weather and climate around the world(Zhang, 2013; Li et al., 2014), such as, the precipitation anomalies in many regions (Mo et al., 2012; Recalde-Coronel et al., 2020); the genesis, intensity, trajectory, and landfall of tropical cyclones (Maloney and Hartmann, 2000; Li and Zhou, 2013); the onset, break and retreat of the Asian summer monsoon (Singh and Bhatla, 2019); and the evolution of El Ni?o events (Lau and Chan 1986; Hendon et al., 2007;Chen et al., 2015).

MJO convection always initiates in the western equatorial Indian Ocean, and then it propagates eastward into the Pacific along the equator at a speed of ～5 m s-1, finally weakening and dying out near the dateline (Hsu et al., 2004).MJO activity mainly occurs over the Indo-Pacific warm pool regions. Therefore, sea surface temperature anomalies(SSTAs) over the Indian and Pacific oceans [e.g., the Indian Ocean Dipole (IOD) and the El Ni?o—Southern Oscillation(ENSO)] play an important role in the variability of the MJO.

The relationship between the MJO and ENSO has been a hot topic since the 1980s and has been widely studied (Li et al., 2014; Chen et al., 2015, 2016). MJO activity is significantly enhanced (weakened) over the equatorial western Pacific before (after) the occurrence of El Ni?o, which reflects the role of MJO in promoting the development of El Ni?o events, as well as the impacts of El Ni?o on MJO activity (Li and Smith, 1995; Chen et al., 2015). In El Ni?o winters, the MJO tends to be a “barotropic” structure in the vertical direction (Li and Smith, 1995), and its eastward propagation is slower and farther east compared with that in La Ni?a winters (Tam and Lau, 2005). The influences of El Ni?o on MJO activity also vary with the diversity of El Ni?o events. MJO activity over the western Pacific is enhanced in the developing stages of eastern Pacific El Ni?o (EP El Ni?o) and is weakened in its mature and decaying stages. Whereas, MJO activity over the western Pacific intensifies in the mature and decaying stages of the central Pacific El Ni?o (CP El Ni?o), and anomalous MJO activity in the developing stages of CP El Ni?o is not prominent(Wang et al., 2015; Chen et al., 2016). The two types of El Ni?o not only affect MJO activity over the western Pacific but also have important impacts on the MJO over the Indian Ocean. Hsu and Xiao (2017) found that moisture and moist static energy in the lower troposphere increase (decrease) during CP El Ni?o (EP El Ni?o), which promotes (inhibits) the development and eastward propagation of the MJO over the Indian Ocean. A recent study indicated that interdecadal variability is also involved in the relationship between the MJO and ENSO, and the enhanced MJO activity associated with the development of EP El Ni?o was more prominent during 1985—2000 (Gushchina, 2019).

The IOD refers to anti-phase variation in SSTAs between the western and eastern tropical Indian Ocean,which has significant impacts on weather and climate (Saji et al., 1999; Saji and Yamagata, 2003). Generally speaking,the westerly in the lower troposphere over the tropical Indian Ocean and the warm SST in the eastern Indian Ocean are favorable for MJO activity and its eastward propagation (Shinoda and Han, 2005). However, these favorable conditions are weakened during the positive phase of the IOD (Shinoda and Han, 2005). In the positive phase of the IOD, negative SSTAs in the eastern Indian Ocean and anomalous easterlies in the lower-level troposphere over the equatorial central Indian Ocean hinder the eastward propagation of the MJO. In contrast, in the negative phase of the IOD, warmer SSTAs in the tropical eastern Indian Ocean and westerly anomalies over the central Indian Ocean promote the development and eastward propagation of the MJO over the Indian Ocean (Seiki et al., 2015). Furthermore, in the negative phase of the IOD, the intensity of lowfrequency (55—100 days) oscillation increases, whereas that of the high-frequency (30—50 days) oscillation decreases significantly, which may result from changes in the background atmospheric circulation and the shallower thermocline ridge and mixed-layer depth in the Indian Ocean (Izumo et al.,2010). MJO activity also modulates the occurrence and development of IOD events through forcing ocean waves (Han et al., 2006). The anomalous easterlies in the suppressed phase of MJO convection can excite upwelling Kelvin waves,which leads to a shallower thermocline in the southeastern Indian Ocean and promotes the development of positive IOD events (Rao et al., 2009). The important role of the MJO in the initiation of the IOD was further confirmed by a numerical study (Yuan and Liu, 2009). The downwelling Kelvin waves induced by the easterly related with the MJO can inhibit the rise of the thermocline in the equatorial central-eastern Indian Ocean, which leads to the termination of the positive phase of the IOD (Rao and Yamagata, 2004).The poleward propagations of the boreal summer intraseasonal oscillations also show distinctive features in the contrasting phases of the IOD. There is a coherent (incoherent) poleward propagation of precipitation over the Indian Ocean during the negative (positive) phase of the IOD (Ajayamohan et al., 2008).

El Ni?o and the IOD represent the SSTAs variability in the tropical Pacific and Indian Oceans, respectively.Recently, many studies have reported that there is an associated mode of SSTAs variation in the tropical Pacific and Indian oceans (Jü et al., 2004; Yang and Li, 2005; Yang et al., 2006). Figure 1 shows the spatial distribution of the first empirical orthogonal function (EOF) mode of the SSTAs in the tropical Pacific—Indian Ocean and its principal component. SSTAs are positive in the equatorial western Indian Ocean and the equatorial central-eastern Pacific, whereas they are negative in the equatorial eastern Indian Ocean and the equatorial western Pacific. SSTAs show a zonal tripole pattern in the tropical Pacific and Indian oceans, which is recognized as the tropical Pacific—Indian Ocean associated mode (PIOAM; Jü et al., 2004; Yang and Li, 2005; Yang et al., 2006). The same mode can also be obtained when the EOF analysis is carried out on tropical subsurface ocean temperature anomalies (Li and Li, 2017; Li et al., 2018).

The PIOAM can simultaneously represent the SSTAs variations in the tropical Pacific and Indian oceans, and it exhibits significant interannual and interdecadal variations(Fig. 1). The PIOAM plays a crucial role in weather and climate anomalies (Yang et al., 2006; Li et al., 2018). Yang and Li (2005) studied the influence of the PIOAM on the South Asian high (SAH) and found that the SAH is weaker(stronger) and is located farther southeast (northwest) in the positive (negative) phase of the PIOAM. Zhou and Mei(2011) indicated that a teleconnection in the Southern Hemisphere, which connects the weather and climate between the tropics and mid—high latitudes, only occurs when the SSTAs in both the Indian and Pacific oceans are taken into consideration simultaneously. Therefore, weather and climate prediction can benefit from an in-depth understanding of PIOAM activity and its impacts (Yang et al., 2006; Li et al., 2018).

Fig. 1. (a) First EOF mode of SSTAs in the tropical Pacific—Indian Ocean and (b) its principal component.

The separate effects of the SSTAs in the tropical Indian and Pacific oceans on MJO activity have been investigated in previous studies, but the effects of the PIOAM, taking the tropical Pacific and Indian oceans as a whole, remain to be investigated. Therefore, we aim to analyze the influences of the PIOAM on MJO activity and explore the possible reasons for these influences from the perspective of atmospheric circulation and moisture budgets.

The remainder of the paper is organized as follows. The datasets and methods used in this study are described in section 2. The impacts of the PIOAM on MJO activity and the reasons for these influences are revealed in sections 3 and 4,respectively. Finally, conclusions and some further discussion are provided in section 5.

2. Data and methods

The reanalysis data used in this paper include the daily specific humidity, sea surface temperature (SST), and winds from ERA-Interim (Dee et al., 2011) with a horizontal resolution of 1.5° × 1.5°; and the daily outgoing longwave radiation (OLR) with a horizontal resolution of 2.5° × 2.5° from NOAA (Liebmann and Smith, 1996). The study period is from 1 January 1979 to 31 December 2018. The climatological mean and long-term linear trend of the reanalysis data were removed, and the MJO signal was obtained using a 30—90-day Lanczos bandpass filter (Duchon, 1979). MJO amplitude refers to the standard deviation of the MJO zonal wind at 850 hPa (MJO OLR) with a three-month sliding window (Hendon et al., 2007).

The prominent interannual variability of the PIOAM occurs in the boreal cold season, and the yearly PIOAM index is defined as the PC1 averaged from September to the following February. The yearly evolution of the PIOAM index from 1979 to 2017 is shown in Fig. 2, which clearly reveals the significant interannual variations of the PIOAM.Anomalous PIOAM years are defined as when the yearly PIOAM index exceeds ±0.85 standard deviations. Based on this criterion, six positive-phase years (1982, 1987, 1991,1997, 2009, and 2015) and seven negative-phase years(1984, 1988, 1998, 1999, 2007, 2010, and 2011) are identified.

3. Impacts of the PIOAM on MJO activity

3.1. Impacts on MJO intensity

Figure 3 shows the composite anomalous amplitude of MJO zonal wind at 850 hPa in the positive and negative phases of the PIOAM. In the positive phase of the PIOAM,amplitudes of MJO zonal wind significantly decrease over the Indian Ocean and Maritime Continent, while they significantly increase over the equatorial central Pacific (Fig. 3a).In the negative phase of the PIOAM, amplitudes of MJO zonal wind increase over the Indian Ocean and Maritime Continent, while they significantly decrease over the equatorial central Pacific (Fig. 3b). The differences in the MJO zonal wind amplitude between the positive and negative phases suggest that MJO activity exhibits significant differences over the tropical Pacific and Indian oceans and shows opposite features between the Indian Ocean and Pacific (Fig. 3c).

Fig. 2. Yearly PIOAM index from 1979 to 2017.

Composite MJO OLR amplitude anomalies in the positive and negative phases of the PIOAM are shown in Fig. 4.In the positive phase of the PIOAM, MJO OLR amplitudes over the Indian Ocean and South China Sea are significantly weakened, while they are significantly enhanced over the equatorial central-eastern Pacific (Fig. 4a). In the negative phase of the PIOAM, MJO OLR amplitudes are enhanced over the equatorial central Indian Ocean, whereas they are significantly decreased over the equatorial centraleastern Pacific (Fig. 4b). The differences in the MJO OLR amplitude between the positive and negative phases indicate that MJO activity exhibits significant differences over the tropical Indian and Pacific oceans (Fig. 4c), which is similar with that for the MJO zonal wind amplitudes.

Fig. 4. As in Fig. 3 but for the anomalous MJO OLR amplitude (units: W m-2).

3.2. Impacts on the eastward propagation of the MJO

Previous studies have reported that SSTAs have important influences on the intensity, velocity, and distance of MJO propagation (Tam and Lau, 2005; Wang et al., 2018).But what are the characteristics of the propagation of MJO during the positive and negative phases of the PIOAM? Figure 5 shows the propagation of MJO OLR averaged over the 10°S—10°N lag regressed onto the Indian Ocean and Pacific MJO indices [presented as the MJO OLR averaged over (10°S—10°N, 70°—90°E) and (10°S—10°N,120°—150°E), respectively]. The MJO shows prominent eastward propagation from the Indian Ocean to the western Pacific under these two conditions. MJO OLR intensity over the Indian Ocean is stronger in the negative phase of the PIOAM, while over the western Pacific to the west of 150°E it is about the same between the positive and negative phases. Another prominent difference between the two phases of the PIOAM is the distance of eastward propagation of the MJO. In the negative phases of the PIOAM,MJO OLR vanishes near 160°E (Figs. 5c and d). However,the MJO propagates to the east of the dateline in the positive phase of the PIOAM (Figs. 5a and b).

The above analyses reveal that the intensity and propagation of the MJO differ significantly between the positive and negative phases of the PIOAM. But what physical processes and mechanisms drive these differences? We explore these issues from the perspective of large-scale circulation and moisture budgets in the next section.

4. Causes of the impacts of the PIOAM on MJO activity

4.1. Large-scale atmospheric circulation

The composite horizontal circulation and specific humidity anomalies at 850 hPa in the anomalous years of the PIOAM are shown in Fig. 6. In the positive phase of the PIOAM, anomalous easterlies prevail over the Indian Ocean and Maritime Continent west of 140°E, but east of 140°E the Pacific is dominated by anomalous westerlies (Fig. 6a).The anomalous specific humidity exhibits a tripole pattern with negative humidity over the eastern Indian Ocean and Maritime Continent and positive humidity over the western Indian Ocean and central-eastern Pacific (Fig. 6a). The characteristics of the abnormal circulation and moisture in the negative phase of the PIOAM are roughly opposite to those in the positive phase (Fig. 6b). There are significant westerly anomalies over the equatorial Indian Ocean and Maritime Continent to the west of 130°E, and there are easterly anomalies to the east of 140°E. The specific humidity is weakened over the western Indian Ocean and central-eastern Pacific, while it is significantly enhanced over the eastern Indian Ocean and Maritime Continent (Fig. 6b).

Fig. 5. Longitude—time diagram of the MJO OLR (units: W m-2) averaged over the 10°S—10°N lag regressed onto the (a, c) Indian Ocean MJO index and (b, d) Pacific MJO index during the (a, b) positive and (c, d) negative phases of the PIOAM. Results passing the significance test at the 90% confidence level are stippled.

Fig. 6. Composite anomalous horizontal winds (vectors; units: m s-1) and specific humidity at 850 hPa (colors; units: g kg-1) in the (a) positive and (b) negative phases of the PIOAM. Only the results that pass the significance test at the 90% confidence level are plotted.

The composite anomalous zonal-vertical circulation and moisture averaged over 10°S—10°N reveal that anomalous ascending motion and positive moisture anomalies occur over the central-eastern Pacific and western Indian Ocean in the positive phase of the PIOAM (Fig. 7a). The Maritime Continent and western Pacific are dominated by negative moisture anomalies and anomalous descending motion. The anomalous fields in the negative phase are roughly opposite to those of the positive phase, but the intensity of the moisture anomalies is slightly weaker. In the negative phase, there is significant anomalous ascending (descending) motion and positive (negative) moisture anomalies over the Maritime Continent and western Pacific (central-eastern Pacific and the western Indian Ocean; Fig. 7b).

Fig. 7. Composite anomalous zonal-vertical circulation (vectors) and specific humidity (colors; units: g kg-1)averaged over 10°S—10°N in the (a) positive and (b) negative phase of the PIOAM. Only the results that pass the significance test at the 90% confidence level are plotted. The vertical velocity has been multiplied by —100 for visualization.

The anomalous ascending (descending) motion and positive (negative) moisture anomalies are (are not) conductive to the generation and maintenance of deep convection(Chen et al., 2016), which is favorable (unfavorable) to MJO activity. Therefore, large-scale atmospheric circulation analysis suggests that the anomalous circulation in the positive phase of the PIOAM is unfavorable to MJO activity over the Maritime Continent and western Pacific, but it is beneficial to MJO activity over the central-eastern Pacific. In the negative phase of the PIOAM, the anomalous circulation is beneficial to MJO activity over the Maritime Continent and western Pacific, but unfavorable to MJO activity over the central-eastern Pacific. These results are consistent with the anomalies of MJO amplitude (Fig. 3).

4.2. Moisture budget

Previous studies have pointed out that the accumulation of moisture in the lower troposphere to the east of the MJO convection favors the generation and propagation of MJO convection (Maloney, 2009; Hsu and Li, 2012, 2014).The intraseasonal moisture tendency equation at a constant pressure level can be written as follows (Hsu and Li, 2012;Chen et al., 2016):

whereq,u,v,ω,Q2andLare the specific humidity, zonal velocity, meridional velocity, vertical velocity, apparent moisture source in the atmosphere, and latent heat of condensation of moisture, respectively. The angle brackets represent the vertical integral from 1000 to 500 hPa, and the primes denote bandpass filtering of 30—90 days. The first three terms on the right-hand side of Eq. (1) are the zonal, meridional, and vertical moisture advection, respectively, which reflect the moisture exchange between the MJO convection and the environment. The fourth term, which mainly reflects the condensation and surface evaporation processes,is the latent heat term (Hsu and Li, 2012; Chen et al., 2016).

Figure 8 shows the MJO circulation and moisture tendency averaged over the 10°S—10°N lag regressed onto the Indian Ocean MJO index. The MJO circulation and moisture tendency are characterized by noticeable eastward propagation during both phases of the PIOAM. During the negative phase of the PIOAM, the MJO moisture tendency and circulation over the Indian Ocean are stronger than those during the positive phase, perhaps causing the difference in MJO intensity over the Indian Ocean. The individual terms of Eq. (1) averaged from day -14 to day -6 and regressed onto the Indian Ocean MJO index are shown in Fig. 9. The difference in the MJO moisture tendencies between the positive and negative phases of the PIOAM is primarily caused by the zonal and vertical moisture advection. To identify the relative contribution of the eddy—eddy and eddy—mean flow interactions, the variations in the zonal and vertical velocities and specific humidity can be divided into a low-frequency background state (LFBS) with a period of greater than 90 days, an MJO component with a period of 30—90 days, and a high-frequency component with a period of less than 30 days, as follows (Hsu and Li, 2012;Chen et al., 2016):

where the overbar, prime, and asterisk denote the LFBS,MJO, and high-frequency components, respectively. Therefore, the moisture zonal and vertical advection can be written as follows:

Fig. 8. MJO zonal-vertical circulation (vectors) and moisture tendency (colors; units: 10-9 kg m-2 s-1) averaged over the 10°S—10°N lag regressed onto the Indian Ocean MJO index in the positive (left-hand panels) and negative (right-hand panels)phases of the PIOAM. Results passing the significance test at the 90% confidence level are stippled for moisture tendency and marked in black for the vectors. The vertical velocity has been multiplied by —100 for visualization.

Fig. 9. Vertically integrated (1000—500 hPa) individual terms for Eq. (1) averaged over (10°S—10°N,70°—90°E), and from day -14 to day -6, regressed onto the Indian Ocean MJO index during the positive(red) and negative (blue) phases of the PIOAM. Units: 10-6 kg m-2 s-1.

The integral zonal and vertical moisture advection terms in the positive and negative phases of the PIOAM averaged from day -14 to day -6 are displayed in Fig. 10. In the negative phase of the PIOAM, the stronger zonal moisture advection is primarily the result of MJO moisture advection by the LFBS zonal wind [the second term on the right-hand side of Eq. (3)] and LFBS moisture advection by the MJO zonal wind [Fig 10a; the fourth term on the right-hand side of Eq. (3)]. The difference in vertical moisture advection is mainly attributable to the LFBS moisture advection by the MJO vertical velocity (Fig 10b). But why is the LFBS moisture advection by the MJO zonal wind stronger in the negative phase of the PIOAM than in the positive phase? In both cases, the low-level easterlies prevail to the east of MJO convection in the Indian Ocean, which is stronger in the negative phase of the PIOAM than that in the positive phase(Figs. 8g and h). Positive (negative) specific humidity over the Maritime Continent (western Indian Ocean) will enhance the LFBS moisture zonal gradient over the Indian Ocean in the negative phase of the PIOAM, while the LFBS moisture zonal gradient in the positive phase of the PIOAM is weakened by the anomalous specific humidity (Fig. 7).Therefore, both the MJO easterlies and zonal gradient of LFBS moisture in the negative phase of the PIOAM are bigger than those in the positive phase, which leads to the stronger LFBS moisture advection.

Moistening over the east of the MJO convection can cause the MJO to propagate farther east (Chen et al., 2016).Figure 11 shows the regressed MJO zonal-vertical circulation and moisture tendency averaged over the 10°S—10°N from day -4 to day 2 onto Pacific MJO OLR index. There is apparent moistening over 160°—180°E in the low-level troposphere during the positive phase of the PIOAM from day —4 to day 2, whereas there is almost no moistening during the negative phase. This significant vertical velocity may lead to an enhanced MJO moisture tendency over the central Pacific during the positive phase of the PIOAM, which allows the eastward propagation of the MJO to reach the international dateline.

Fig. 10. As in Fig. 9 but for the individual terms of the moisture zonal and vertical advection. Units:10-6 kg m-2 s-1.

The regressed moisture budget terms averaged over the central Pacific from day -4 to day 5 are displayed in Fig. 12.It shows that the MJO moisture tendency is positive in the positive phase of the PIOAM, whereas it is negative in the negative phase. The difference in the moisture tendencies during the two phases is mainly attributable to the vertical and meridional moisture advection. The meridional and vertical moisture advection can be decomposed into nine terms according to Eq. (4). Figure 13 shows the integral meridional and vertical moisture advection terms in the positive and negative phases of the PIOAM averaged from day -4 to day 5. The meridional moisture advection is mainly attributable to LFBS moisture advection by the MJO meridional wind (Fig. 13a). In addition, the LFBS moisture advection by the LFBS meridional wind is also significantly different during the positive and negative phase of the PIOAM. The LFBS moisture advection by the MJO vertical velocity causes the difference in vertical moisture advection during positive and negative phases of the PIOAM (Fig. 13b).

Fig. 11. MJO zonal-vertical circulation (vectors) and moisture tendency (colors; 10-9 kg m-2 s-1) averaged over the 10°S—10°N lag regressed onto the Pacific MJO index in the positive (left-hand panels) and negative (right-hand panels) phases of the PIOAM. Results passing the significance test at the 90% confidence level are stippled for moisture tendency and marked in black for the vectors. The vertical velocity has been multiplied by —100 for visualization.

Fig. 12. Vertically integrated (1000—500 hPa) individual terms of Eq. (1) averaged over (10°S—10°N,160°—180°E), and from day -4 to day 5, regressed onto the Pacific MJO index during the positive(red) and negative (blue) phases of the PIOAM. Units: 10-6 kg m-2 s-1.

5. Conclusions and discussion

The PIOAM represents coordinated variation of SST in the Pacific and Indian oceans, and it can modulate the atmospheric systems in both the tropics and extratropics. Our analyses indicate that the PIOAM has significant effects on the intensity and propagation of MJO activity. The amplitudes of MJO zonal wind and OLR are weakened over the Indian Ocean and enhanced over the central-eastern Pacific during the positive phase of the PIOAM. Whereas, they are enhanced over the Indian Ocean and weakened over the central-eastern Pacific during the negative phase of the PIOAM.The MJO exhibits prominent eastward propagation from the Indian Ocean to the western Pacific in anomalous PIOAM years. However, the MJO can propagate farther east in the positive phase of the PIOAM. The eastward propagation of the MJO extends to east of the dateline during the positive of the PIOAM, but it is mainly confined to west of 160°E during the negative phase.

Fig. 13. As in Fig. 12 but for the individual terms of moisture meridional and vertical advection.Units: 10-6 kg m-2 s-1.

Large-scale atmospheric circulation and moisture are significantly different between the two phases of the PIOAM,which causes the different features of MJO activity. During the positive phase of the PIOAM, anomalous easterlies occur over the Indian Ocean and anomalous westerlies occur over the central-eastern Pacific, and anomalous descending motion and negative specific humidity occur over the Maritime Continent, and anomalous ascending and positive specific humidity occur over the Indian Ocean and centraleastern Pacific. Therefore, the large-scale circulation and moisture anomalies provide a favorable (unfavorable) background state for the activity and propagation of the MJO over the central Pacific (Indian Ocean and Maritime Continent). The large-scale circulation and moisture anomalies in the negative phases of the PIOAM are almost opposite to those in the positive phase, which is (is not) conducive to the activity of the MJO over the Indian Ocean (central Pacific).

Moisture budget analysis indicates that the stronger moistening over the Indian Ocean in the negative phase of the PIOAM induces stronger MJO activity. This moistening is mainly attributable to the zonal and vertical moisture advection. The zonal moisture advection is driven by the LFBS moisture advection by the MJO zonal wind and MJO moisture advection by the LFBS zonal wind. The vertical moisture advection is mainly attributable to the LFBS moisture advection by the MJO vertical velocity. The MJO moisture budget over the central Pacific was further analyzed. Results suggest that the stronger moistening over the central Pacific in the positive phase of the PIOAM leads to farther east propagation of the MJO over the western Pacific. This moistening is mainly attributable to the vertical and meridional moisture advection. The meridional moisture advection is driven by the LFBS moisture advection by the MJO zonal wind. Vertical moisture advection is mainly attributable to the LFBS moisture advection by the MJO vertical velocity. Both the moisture and the circulation intensity differ significantly during the two PIOAM phases, which leads to differences in moistening over the Indian Ocean and central Pacific. Thus, the MJO activity over the Indian Ocean during the negative phase of the PIOAM is more robust, and the MJO over the Pacific during the positive phase of the PIOAM propagates farther east.

Fig. 14. Longitude—time diagram of the MJO OLR (units: W m-2) averaged over the 10°S—10°N lag regressed onto the Pacific MJO index during (a) El Ni?o, (b) La Ni?a, (c) IOD positive phase, (d) IOD negative phase, (e) PIOAM positive phase, and (f) PIOAM negative phase. Results passing the significance test at the 90% confidence level are stippled.

ENSO and the IOD also influence MJO activity by modulating circulation and moisture (Shinoda and Han, 2005;Chen et al., 2015, 2016). To compare the effects of the PIOAM on the MJO with those of ENSO and the IOD, we define ENSO and IOD years based on the oceanic Ni?o index (ONI) and IOD index following the method of Trenberth (1997) and Yuan and Liu (2009). El Ni?o (La Ni?a) events are defined as having an ONI index exceeding+0.5°C (-0.5°C) for at least five consecutive months(Trenberth, 1997). There were 12 El Ni?o (1982, 1986,1987, 1991, 1994, 1997, 2002, 2004, 2006, 2009, 2014, and 2015) and 10 La Ni?a events (1983,1984,1988, 1995, 1998,1999, 2000, 2007, 2008, and 2010) from 1979 to 2018. The IOD events were selected if the IOD index during the Autumn exceeded 0.85 standard deviations (Yuan and Liu,2009). Based on this definition, seven positive phases(1982, 1994, 1997, 2006, 2011, 2012, and 2015) and seven negative phases (1981, 1990, 1992, 1996, 1998, 2010, and 2016) of the IOD were identified. The MJO OLR averaged over 10°S—10°N lag regressed onto the Pacific MJO index during the PIOAM, ENSO and IOD events is shown in Fig. 14.MJO OLR is comparable over the Indian Ocean between El Ni?o and La Ni?a events, but MJO OLR is stronger over the Pacific and propagates farther east in El Ni?o winter(Figs. 14a and b). The significant MJO OLR can propagate near 140°W in El Ni?o events, while it dies out near 180° in La Ni?a events. MJO OLR is stronger over both the Indian Ocean and Pacific and can propagate farther east in the negative IOD events compared with those in the positive IOD events (Figs. 14c and d). MJO activity in the PIOAM significantly differs from that during the ENSO and IOD events(Figs. 14e and f). The MJO OLR over the Indian Ocean is stronger in the negative phases of the PIOAM than that in the positive phases, which is similar to what occurs in anomalous IOD events. Meanwhile, the MJO OLR can propagate farther east in the positive phase of the PIOAM, which is similar to what occurs during the ENSO events and differs from that during the IOD phases. Significant MJO activity in the positive phase of the PIAOM is mainly confined to west of 180°, while it is also prominent to the east of 180° in the warm phase of ENSO. Therefore, the impacts of the PIOAM on MJO activity are different from those of ENSO and the IOD. The coincident variation in the SSTAs in the Indian Ocean and Pacific should be considered when predicting the activity of the MJO.

Acknowledgements.We thank the anonymous reviewers for their careful comments and suggestions. This work was supported by the National Key Research and Development Program of China(Grant No. 2018YFC1505901) and the National Natural Science Foundation of China (Grant Nos. 41605051, 41520104008,41475070 and 41575062).