J. Meteor. Res.  2017, Vol. 31 Issue (4): 665-677 PDF
http://dx.doi.org/10.1007/s13351-017-6178-3
The Chinese Meteorological Society
0

#### Article Information

Qi XU, Zhaoyong GUAN . 2017.
Interannual Variability of Summertime Outgoing Longwave Radiation over the Maritime Continent in Relation to East Asian Summer Monsoon Anomalies. 2017.
J. Meteor. Res., 31(4): 665-677
http://dx.doi.org/10.1007/s13351-017-6178-3

### Article History

in final form March 23, 2017
Interannual Variability of Summertime Outgoing Longwave Radiation over the Maritime Continent in Relation to East Asian Summer Monsoon Anomalies
Qi XU, Zhaoyong GUAN
1. Key Laboratory of Meteorological Disaster, Ministry of Education/Joint International Research Laboratory of Climate and Environment Change/Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science & Technology, Nanjing 210044
ABSTRACT: The Maritime Continent (MC) is under influences of both the tropical Pacific and the Indian Ocean. Anomalous convective activities over the MC have significant impacts on the East Asian summer monsoon (EASM) and climate in China. In the present study, the variation in convective activity over the MC in boreal summer and its relationship to EASM anomalies are investigated based on regression analysis of NCEP–NCAR reanalysis and CMAP [Climate Prediction Center (CPC) Merged Analysis of Precipitation] data, with a focus on the impacts of ENSO and the In-dian Ocean Dipole (IOD). The most significant interannual variability of convective activity is found over 10°S–10°N, 95°–145°E, which can be roughly defined as the key area of the MC (hereafter, KMC). Outgoing longwave radiation anomaly (OLRA) exhibits 3-to 7-yr periodicities over the KMC, and around 70% of the OLRA variance can be explained by the ENSO signal. However, distinct convection and precipitation anomalies still exist over this region after the ENSO and IOD signals are removed. Abnormally low precipitation always corresponds to positive OLRA over the KMC when negative diabatic heating anomalies and anomalous cooling of the atmospheric column lead to abnormal descending motion over this region. Correspondingly, abnormal divergence occurs in the lower troposphere while convergence occurs in the upper troposphere, triggering an East Asia–Pacific/Pacific–Japan (EAP/PJ)-like anomalous wave train that propagates northeastward and leads to a significant positive precipitation anomaly from the Yangtze River valley in China to the islands of Japan. This EAP/PJ-like wave pattern becomes even clearer after the removal of the ENSO signal and the combined effects of ENSO and IOD, suggesting that convective anomalies over the KMC have an important impact on EASM anomalies. The above results provide important clues for the prediction of EASM anomalies and associated summer precipitation anomalies in China.
Key words: Maritime Continent     convective activity     ENSO     Indian Ocean Dipole (IOD)     East Asian summer monsoon (EASM)
1 Introduction

Ramage (1968) defined the geographic area over 10°S–20°N, 90°–150°E as the Maritime Continent (MC) and discussed its impact on atmospheric circulation. The MC consists of many islands and shallow seas. Situated within the tropical warm pool that extends from the tropical western Pacific to the Indian Ocean, the MC is collocated with the tropical Pacific and the Indian Ocean, and behaves as a linkage between the East Asian monsoon and the Australian monsoon regions. The MC region experiences a typically maritime climate, with multi-scale spatial and temporal climate variability and close associations with, for example, the Madden–Julian Oscillation (MJO; Madden and Julian, 1972; Oh et al., 2013), ENSO, Indian Ocean Dipole (IOD), intertropical convergence zone (ITCZ), South Pacific convergence zone (SPCZ), Walker circulation, and monsoons. Due to its special geographic location and climate background, further in-depth study of the climate anomalies over the MC region is necessary.

Climate variability in the MC region and its possible impact are closely related to local convective activity. Previous studies have indicated that strong convective activity occurs frequently over the MC region, with large diurnal variation, which plays an important role in both monsoon circulation changes and global climate variability (Neale and Slingo, 2003; Mori et al., 2004; Teo et al., 2011; Wang et al., 2016). Different to that in the extratropical region, convective activity in the tropics is often accompanied by large amounts of latent heat release and the upward transport of water vapor. Many factors can affect convective activity over the MC region. Some researchers have argued that radiative forcing is the primary reason for the diurnal variation of convection, and water vapor convergence is just a secondary reason (Liu and Moncrieff, 1998); while others have proposed that the land–sea temperature contrast and land–sea breeze are the most important reason, since abnormal convective activity is easily triggered when large-scale circulation affects sea surface temperature (SST) and humidity (Araki et al., 2006). In addition, convective activity over the MC region is also closely linked with the eastward propagation of the MJO and inhomogeneous land–sea distribution (Shibagaki et al., 2006; Wu and Hsu, 2009; Zhou and Murtugudde, 2010; Oh et al., 2012; Li, 2014).

Climate anomalies over the MC region are highly correlated with ENSO and IOD (Saji et al., 1999; Ashok et al., 2001; Qian et al., 2012). In fact, SST and precipitation over the MC region are highly responsive to ENSO and IOD. ENSO can greatly affect precipitation in the MC region during June–August (dry season) (Chang et al., 2004). When El Niño occurs, SST anomalies (SSTAs) in the warm pool region trigger abnormal easterlies in eastern Indonesia. As a result, the SST becomes cooler than normal and precipitation decreases (Hackert and Hastenrath, 1986; Hendon, 2003). Similarly, climate anomalies over the MC region can affect ENSO and IOD. For example, Huang (1990) pointed out that zonal wind anomalies in the equatorial tropics could provoke equatorial ocean waves, which is an important dynamic factor that promotes the occurrence of ENSO. In the positive phase of the IOD, convection in the equatorial Indian Ocean is more frequent than that over the western coast, and precipitation decreases over the MC region. Several previous studies have shown that both the Indonesian throughflow and the local air–sea feedback off Sumatra are possible triggering factors for the IOD (Baquero-Bernal et al., 2002; Song and Gordon, 2004). In the tropics, the Pacific Ocean and the Indian Ocean interact with each other via the "atmospheric bridge" (Alexander et al., 2002), while changes in the Walker circulation play an important role. The ascending branch of the Walker circulation is exactly located over the MC region. Meng and Wu (2000) investigated the interaction between the Pacific Ocean and the Indian Ocean, and proposed the concept of gearing between the Indo-monsoon Circulation and the Pacific–Walker Circulation (GIP). ENSO events correspond well to interannual GIP anomalies, that is, the gearing operates in opposite directions in response to cold and warm events, respectively. The engagement point of abnormal gearing is located near the islands of Indonesia in the MC region.

Climate variability and change in the MC region often lead to anomalies in the East Asian summer monsoon (EASM) (Wang et al., 2015). Tao and Chen (1987) revealed that the outbreak of the EASM first occurs in the northern South China Sea, and then extends northward to reach mainland China, the Pacific Ocean, and to the south of Japan. Ding et al. (2004) suggested that the South China Sea is the primary source of water vapor for the EASM's precipitation, and the intensity of the South China Sea monsoon is highly correlated with precipitation over China: a strong (weak) monsoon often leads to lower (higher) than normal precipitation over the upper and lower reaches of the Yangtze River, and higher (lower) than normal precipitation over North China. The large heat capacity in the western Pacific warm pool can intensify convection in the Philippines region, leading to propagation of quasi-geostrophic planetary waves and a northward shift of the western Pacific subtropical high in summer. The East Asia–Pacific (EAP) teleconnection pattern subsequently forms; meanwhile, the EASM circulation intensifies (Wu et al., 2013; Chen and Zhou, 2014). The IOD also affects the EASM to a certain degree. The IOD can induce vorticity anomalies over the Bay of Bengal and the South China Sea, and anomalous adiabatic heating over India. Through a triangular teleconnection pattern, the IOD affects the EASM (Guan and Yamagata, 2003). In the positive phase of the IOD, cyclonic circulation anomalies occupy southern China and western Pacific, while anticyclonic circulation anomalies prevail over Northeast China, the Korean Peninsula, and Japan, and the northward propagation of the EASM weakens.

In general, the weather and climate variability over the MC region is highly correlated with the EASM and climate anomalies in China. Specifically, the relation of convective activity over the MC and the EASM varies in different areas within the MC. What are the characteristics of the changes in convective activity within the MC region, and where does the strongest variation occur? Generally, the ENSO signal is most distinct in the Northern Hemisphere in winter, but it still exists in summer. The correlation coefficient between Niño3 (Niño3.4) in winter and in summer is 0.80 (0.86), indicating that the ENSO signal is characterized by strong persistence. Therefore, is the outgoing longwave radiation (OLR) variability in summer over the largest variability area of the MC region still significant after the removal of the ENSO and IOD signals? What is the relationship be-tween the OLR variability and EASM changes? This study seeks to answer these questions based on analysis of the OLR variability over the MC region.

As mentioned above, the relationships between convective activity in various areas within the MC and the EASM are different. This study aims to find where the strongest variation of convective activity is, within the MC region. Given the ENSO signal's strong persistence in the MC, it is important to determine whether the convective activity over the identified area is still significant after the removal of the ENSO and IOD signals, and its relation with the EASM changes. The results are helpful for better understanding the mechanisms of the EASM anomalies, and may provide important clues for prediction of summer precipitation anomalies in China.

2 Data and method 2.1 Data

The NCEP–NCAR reanalysis (Kalnay et al., 1996; Kanamitsu et al., 2002) of monthly mean winds, temperature, surface pressure, sensible heat flux, mixing ratio of water vapor, latent heat flux, SST, and OLR are used in this study. The horizontal resolution of the data is 2.5° (latitude) × 2.5° (longitude). Surface air temperature at 2 m is on a global Gaussian grid. The CMAP [Climate PredictionCenter (CPC) Merged Analysis of Precipitation] (Xie and Arkin, 1997) monthly mean precipitation is also used in this study. The indices of Niño3.4, Niño4, and Niño1+2, as well as the Dipole Mode Index, are taken from the NOAA's Earth System ResearchLaboratory (https://www.esrl.noaa.gov/psd/gcos_wgsp/Timeseries/). The data cover all the summers between 1979 and 2013. The summer refers to June–September, and the summer average are defined as the average of the data over June–September.

2.2 Method

ENSO and IOD have significant influences on meteorological variables in the tropics (e.g., Alexander et al., 2002; Ashok et al., 2003; Saji and Yamagata, 2003; Jin et al., 2016). Even in the tropical Indo-Pacific sector, a combined SSTA mode exists that involves both ENSO and IOD signals (Yang et al., 2006). In order to reveal the nature of the climate over the MC region with and without the influences of ENSO and IOD and its relationship with climate in East Asia in this study, the (partial) linear regression method is employed to filter out ENSO and IOD signals. The F-test is then employed to examine the statistical significance of the regression analysis results. The Morlet wavelet method is utilized to reveal the periodic cycles of the time series (Torrence and Compo, 1998).

In order to further analyze the relationship between convective activity in the MC region and the East Asian monsoon climate, we compute the apparent atmospheric heat source $\langle {Q_1}\rangle$ and the apparent moisture sink $\langle {Q_2}\rangle$ (Yanai et al., 1973; Luo and Yanai, 1984), which can be expressed as:

 ${Q_1} = {c_p}(\frac{{\partial T}}{{\partial t}} + {V} \cdot \nabla T + {(\frac{p}{{{p_{_0}}}})^κ}\omega \frac{{\partial \theta }}{{\partial p}}),$ (1)
 $\!\!\!\!\!\!\!\!\!\!\!\!{Q_2} = - L(\frac{{\partial q}}{{\partial t}} + {V} \cdot \nabla q + \omega \frac{{\partial q}}{{\partial p}}).$ (2)

The three terms on the right-hand side of the above two equations represent local change, horizontal advection, and vertical advection, respectively. Now, using ${\langle ^{}}\;\;\rangle = \displaystyle\frac{1}{g}\int_{{p_{\rm t}}}^{{p_{\rm s}}} {{\!\!(^{}}\;\;)} {\rm dp}$ to denote the vertical integration of a term in the above two equations, we have

 $\langle {Q_1}\rangle = ({ LP_{\rm r}} + { LC} - {LE}) + {Q_{\rm s}} + \langle {Q_{\rm r}}\rangle ,$ (3)
 $\!\!\!\!\!\!\!\!\!\!\!\!\!\!\langle {Q_{\rm{2}}}\rangle = ({LP_{\rm r}} + {LC }-{LE}) - {{LE}_{\rm s}},$ (4)

and

 $\Delta Q = \langle {Q_{\rm r}}\rangle + \left( {{Q_{\rm s}} + {LE_{\rm s}}} \right),$ (5)

where L is the latent heat of condensation, Pr is the amount of precipitation, C is the total amount of liquid water from water vapor condensation, E is the evaporation of cloud drops in the column, Es is the evaporative surface water vapor flux, Qs is the surface sensible heat flux, $\langle {Q_{\rm r}}\rangle$ is the vertically integrated radiative heating or cooling, ps is the surface pressure, and pt is the pressure at the top (pt = 300 hPa). The physical quantity ∆Q is the difference between $\langle {Q_1}\rangle$ and $\langle {Q_2}\rangle$ (the former minus the latter), which represents the net diabatic heating rate, without the latent heat release inside the air column.

3 OLR variability in the MC region and definition of the key area of the MC 3.1 Variability of OLR in the MC region

Due to its geographic location and topography, convective activity in the MC region is relatively strong, while synoptic weather systems often develop and disappear rapidly. The OLR can represent the intensity of a local convective system well. Figures 1a and 1b display the multi-year climatological mean OLR and root-mean-square (RMS) of OLR anomalies in the Northern Hemisphere in summer (June–September) over the MC region, respectively. Overall, the climatological mean OLR in the MC region decreases from south to north, indicating weak convective activity in the south and strong convective activity in the north. Two low OLR centers are located over western Indonesia and the Philippines, respectively, where convection is active. A large OLR center is found over the south of the islands of Indonesia (around 120°E in the Southern Hemisphere). Figure 1b shows large RMS values of OLR between 10°S and 10°N, indicating a strong temporal variability of convective activity in this region, despite convection in this region being weaker overall than in the north. Based on these results, we roughly define the area (10°S–10°N, 95°–145°E) as the key area of the MC region (KMC). This region covers most of Indonesia, is symmetric about the equator, and has a large interannual variability of OLR.

 Figure 1 (a) Multi-year mean OLR (W m–2) over the MC region. (b) RMS (root-mean-square) of original OLR anomalies. (c) RMS of OLR anomalies after the removal of the ENSO signal. (d) RMS of OLR after the removal of both the ENSO and IOD signals. The rectangular area with a green frame denotes the KMC region.

Previous studies have shown that ENSO and IOD have important impacts on climate variability over the MC region (Ashok et al., 2001; Qian et al., 2012; Huo and Jin, 2016; Jin et al., 2016). Therefore, we ask the question: what is the nature of the change in OLR in the MC region after the ENSO and IOD signals have been removed?

In this study, the linear regression method is utilized to remove the ENSO signal represented by the Niño3.4 index, and the result is shown in Fig. 1c. It is clear that large OLR interannual variability still exists in the KMC region. Comparing Fig. 1b with Fig. 1c reveals that the strongest ENSO impact occurs over 5°–0°S, 110°–130°E, suggesting that the concurrent impact of the equatorial Pacific SSTA is the largest over the equatorial area between 110° and 130°E, and relatively small in other regions. On the basis of these data, the linear regression method is employed again to remove the IOD signal after the ENSO signal has been removed from the IOD index in advance. The result is displayed in Fig. 1d. By examining Figs. 1d, c together, we can see that the IOD impact largely focuses on 10°–0°S, 90°–110°E, which is coincident with the equatorial southwestern Indian Ocean pole of the IOD. This result is consistent with the fact that the IOD is a coupled ocean–atmosphere phenomenon in the equatorial Indian Ocean, independent of ENSO (Saji et al., 1999; Ashok et al., 2001).

Since OLR in the KMC is closely related to ENSO and IOD, we further explore the climatological features of OLR in the KMC and its possible links with ENSO, IOD, and EASM.

3.2 Periodicities in KMC convection

For convenience, OLRM is used to represent multi-year-averaged summertime (June–September) OLR, and OLRA stands for OLR anomaly. Square brackets indicate the average over the KMC. Therefore, we have:

 ${I_{\rm a}} = [\rm OLR] - [\rm OLRM].$ (6)

The linear regression method is employed to filter out ENSO signals from Ia, that is,

 ${I_{\rm b}} = {I_{\rm a}} - \alpha {I_{\rm Niño3.4}}.$ (7)

After the Niño3.4 signals have been removed from the IOD index (IIOD) by linear regression, we obtain the modified IOD index ImIOD, which is then used to filter out IOD signals from Ib by regression. This can be expressed as:

 ${I_{\rm c}} = {I_{\rm b}} - \beta {I_{\rm mIOD}}.$ (8)

Here, The variables with subscripts "a", "b", and "c" stand for the original, ENSO-removed, and both ENSO and IOD removed time series. In order to better display the relationships of OLR in the KMC with ENSO and IOD, we list the correlation coefficients of the OLRA time series with that of the IOD and various ENSO indices in Table 1. The ENSO indices of ICP and IEP, for CP (central Pacific)-and EP (eastern Pacific)-type ENSO, are obtained based on the summertime SSTA by the joint regression analysis of Kao and Yu (2009).

Convective activity in the KMC is more closely correlated with CP-type ENSO than with EP-type ENSO. The correlation coefficient between Ia and Niño3.4 index reaches 0.84, while that between Ia and IOD index reaches 0.59. Apparently, ENSO and IOD both play an important role in weather and climate variations over the KMC region.

Table 1 Correlation coefficients between OLR anomalies and various SST indices. The critical value for the 95% confidence level is found to be 0.33
 Index INiño3.4 INiño1+2 ICP IEP IIOD ImIOD Ia 0.84 0.51 0.64 0.33 0.59 0.49 Ib 0.00 0.06 –0.05 0.07 0.44 0.48 Ic 0.00 –0.08 0.08 –0.06 0.00 0.00
 Figure 2 Time series of original OLRA (OLR anomaly; Ia, black bars), OLRA with the ENSO signal removed (Ib, red line), and OLRA with both the ENSO and IOD signals removed (Ic, blue line), averaged over the KMC region.

The OLRA in the KMC region demonstrates a significant interannual variability. However, after the ENSO and IOD signals have been filtered out by regression analysis, we find that the amplitude of the OLRA variability distinctly decreases. As shown in Fig. 2, the OLRA variability amplitude changes significantly before and after the ENSO signal is removed, particularly in 1982, 1987, 1988, 1997, 1999, and 2010. As we know, strong El Niño events occurred in the years of 1982, 1987, and 1997, while strong La Niña events occurred in 1988, 1999, and 2010. El Niño (La Niña) events typically peak in strength in winter (early spring) and the subsequent summer, and SSTs are abnormally low (high) in the equatorial western Pacific, including the MC region. As a result, the sensible heat flux transfer from the ocean up into the atmosphere weakens (intensifies), convection is relatively weak (strong), and OLR is higher (lower) than normal. These results are consistent with the results of some previous studies (e.g., Torrence and Webster, 1999; Mc-Bride et al., 2003). The ratio of the variance of Ib to that of Ia is 29.35%, which indicates that the contribution of ENSO accounts for 70.65% of the OLRA variance. This explains why the amplitude of the interannual OLRA variability significantly decreases after the ENSO signal is removed.

Comparing Ib (with ENSO signal removed) with Ic (with both ENSO and IOD signals removed) reveals no large difference between them. This indicates that, despite the abnormally low (high) SST in the equatorial eastern Indian Ocean, weaker (stronger) than normal latent and sensible heat flux exchanges between the ocean and atmosphere, relatively weak (strong) convection, and anomalously higher (lower) OLR in the years of positive-phase (negative-phase) IOD, the overall impact of pure IOD on the OLRA is not significant. Therefore, ENSO affects convective activity over the KMC region mainly via abnormal oceanic and atmospheric circulations during ENSO events. In some strong ENSO years (e.g., 1987, 1988, 1999, and 2010), even the OLRA sign was changed due to ENSO impacts. Compared with the impacts of ENSO, the IOD impacts on OLRAs are generally much smaller.

 Figure 3 Morlet wave power spectra of the OLRA averaged over the KMC region: (a) the original OLRA (Ia); (c) the OLRA after removal of the ENSO signal (Ib); and (e) the OLRA after removal of both the ENSO and IOD signals (Ic). Shaded areas indicate statistical significance at the 95% confidence level. The cone-shaped spectra represent the demarcation for areas affected by the border effects of the wavelet window. Panels (b), (d), and (f) are the global power spectra with red-noise checking at the 95% confidence level for Ia, Ib, and Ic, respectively.

The periodicities of OLRA variation in the KMC are complicated, but generally present a 3–7-yr cycle. Figure 3 displays the results of the Morlet wavelet analysis of Ia, Ib, and Ic. Before the ENSO and IOD signals are filtered out, the OLRA in the KMC region presents a 4–5-yr cycle from the 1980s to around 2000. A 2–5-yr cycle is particularly prominent in the 1990s. These cycles are similar to the 3–7-yr cycle of ENSO (Torrence and Webster, 1999) and the 4–5-yr and quasi-2-yr cycles of the IOD (Ashok et al., 2001). After the ENSO signal is removed, the OLRA shows a 3-and a 6–7-yr cycle before 2000 and a 4-yr cycle after 2000. After both the ENSO and IOD signals are removed, periodic quasi-3-and quasi-8-yr changes dominate the OLRA variations in the 1980s, while a periodic 5–7-yr change is more distinct in the 1990s.

It is interesting that the quasi-3-yr periodic cycle is distinct in the OLRA power spectra before (Fig. 3b) and after (Fig. 3d) the ENSO signal is removed, and after both the ENSO and IOD signals are removed (Fig. 3f), over the MC region. However, a 7-yr cycle is more distinct after the ENSO and IOD signals are removed, indicating that ENSO and IOD mainly affect periodic convective activities with a cycle of less than 7 yr in the KMC region. The above result also indicates that convective activity in the KMC region largely varies on significant quasi-3-and quasi-7-yr cycles, and such convection bears no close relation with ENSO or IOD.

4 Relationship between OLRA and diabatic forcing in KMC

Convective activity in the KMC region is associated with both the remote forcing of SSTAs in the equatorial central–eastern Pacific and the local diabatic forcing. The correlation coefficients of the time series of domain-averaged $\langle {Q_1}\rangle$ and ∆Q with OLRA in the KMC region are computed (Table 2). The results show that the OLRA variation is always negatively correlated with the dia-batic heating, regardless of whether ENSO and/or IOD influences exist. That is, the stronger the diabatic heating, the more intense the convection in the KMC region.

Table 2 Correlation coefficients between OLRA and area-averaged anomalous diabatic heating rate in the KMC region. The critical value for the 95% confidence level is 0.33
 Time series ${\langle {Q_1}\rangle _{\rm a}}$ ${\langle {Q_1}\rangle _{\rm b}}$ ${\langle {Q_1}\rangle _{\rm c}}$ ∆Qa ∆Qb ∆Qc Ia –0.83 –0.29 –0.27 –0.57 –0.19 –0.09 Ib –0.35 –0.54 –0.50 –0.30 –0.35 –0.17 Ic –0.37 –0.57 –0.57 –0.15 –0.17 –0.19

Before the ENSO and IOD signals are removed, the correlation coefficient between Ia and ${\langle {Q_1}\rangle _{\rm a}}$ is –0.83, and that between Ia and ΔQa reaches –0.57. Since ΔQa is the sum of the net radiative heating rate anomaly and surface heat flux anomaly, it is apparent that abnormal heating of the atmospheric column can directly intensify convection.

After the ENSO signal is removed, the correlation coefficient between Ib and ${\langle {Q_1}\rangle _{\rm b}}$ reaches –0.54, while that between Ib and ΔQb can reach –0.35. This suggests that ENSO plays an important role in the above relationship, but it cannot change the basic linkage between convection and diabatic heating of the air column.

When both the IOD and ENSO signals are removed, a close relationship between Ic and ${\langle {Q_1}\rangle _{\rm c}}$ still exists; the correlation coefficient between Ic and ${\langle {Q_1}\rangle _{\rm c}}$ is –0.57. However, the correlation coefficient between Ic and ΔQc is only –0.19. This indicates that the abnormal heating of the atmospheric column exclusive of latent heat release can exert certain impacts on convection in the KMC region, but the impacts are not strong.

To further examine the relationship between OLRA and abnormal diabatic heating rate, linear regressions of the apparent heat source $\langle {Q_1}\rangle$ and ΔQ on the time series of Ia, Ib, and Ic during boreal summer are computed. The results are shown in Fig. 4.

The significant negative correlation between the area-averaged OLRA and diabatic heating in the KMC region suggests that the diabatic heating has critically important impacts on convective activity in this region. Before the ENSO signal is removed (Fig. 4a), large negative $\langle {Q_1}\rangle$ anomalies are found over most of the KMC region. An abnormal heating center is located over the warm pool region of the western Pacific to the northeast of the KMC, while negative heating anomalies occur to the east of the Taiwan island and Japan. This pattern is consistent with the EAP/PJ (Pacific–Japan) teleconnection pattern (Huang, 1987; Nitta, 1987). Large positive heating anomalies occur in Indochina and southern China, which are possibly associated with the Gill-type response (Gill, 1980). Figure 4b shows that negative ΔQ anomalies are dominant in the KMC region, while positive heating anomalies occur to the east of the KMC and negative heating anomalies appear in the South China Sea, the Philippines, and the Pacific Ocean to the east of the Taiwan island. Positive heating anomalies can also be found in Indochina. Comparing Fig. 4a with Fig. 4b reveals that the negative heating anomaly center in the KMC decreases significantly after the latent heating anomaly is removed, indicating that latent heat release is important for convective activity in the KMC region. Meanwhile, negative ΔQ anomalies suggest that the radiative heating and sensible heating anomalies in the atmospheric column also affect convective activity. After the ENSO signal is removed, the negative $\langle {Q_1}\rangle$ anomaly center (Fig. 4c) in the central KMC disappears, a large negative heating anomaly appears in the Philippines, and the positive heating anomaly center in the warm pool of the western Pacific weakens and shifts eastward. Negative ΔQ anomalies in the central KMC (Fig. 4d) disappear, the negative center near the Philippines weakens, and the heating anomalies in the warm pool of the western Pacific become insignificant. Apparently, the ENSO impacts on the KMC and MC region concentrate on the warm pool area of the western Pacific and nearby 120°E. After both the ENSO and IOD signals are removed, negative anomalies of $\langle {Q_1}\rangle$ (Fig. 4e) and ΔQ (Fig. 4f) weaken greatly in the southwestern KMC, but change little in other areas.

 Figure 4 Spatial patterns of regression coefficients for the regression of (a, c, e) $\langle {Q_1}\rangle$ and (b, d, f) ΔQ on the time series of (a, b) Ia, (c, d) Ib, and (e, f) Ic. The stippled areas indicate the 90% confidence level with the F-test.
5 Relationship between OLRA in KMC and the EASM 5.1 Relationship with changes in circulation

Heating anomalies in the KMC region induce large-scale anomalous convective activity, which may subsequently affect the EASM circulation. On the interannual timescale, ENSO and IOD are two strong signals to the east and west of the KMC, respectively, which affect circulation changes in the KMC region. However, after the ENSO and IOD signals are removed, circulation responses still exist in the KMC region. Outputs produced from the regression of circulation during 1979–2013 on the time series of Ia, Ib, and Ic as seen in Fig. 2, including the original series, the series with the ENSO signal removed, and the series with both the ENSO and IOD signals removed, are presented in Fig. 5.

Convective activity in the KMC region is closely linked with the EASM circulation. Figure 5a shows that westerly anomalies at 850 hPa are dominant in the KMC region, which are associated with the positive SSTAs in the equatorial central–eastern Pacific in boreal summer (Meng and Wu, 2000). Easterly anomalies are significant in the tropical Indian Ocean to the southwest of the KMC, which are favorable for the formation of the IOD (Saji et al., 1999). Apparently, significant divergent flows can be found in the KMC region. It is observed that most of China is under the control of the large anomalous cyclonic circulation centered in northwestern Pacific. The formation of this large cyclonic circulation is largely attributable to the Gill-type response (Gill, 1980) of the atmosphere to warm SSTAs in the equatorial central Pacific, where strong convergence is generated and compensated partly by the anomalous divergence in the KMC region. At 200 hPa (Fig. 5b), convergence can be found in the KMC region, which is responsible for the descending motion in this region and ascending motion in southern China, leading to local increases in precipitation.

After the ENSO signal is removed, the circulation anomalies in the KMC region that are associated with the OLRAs demonstrate a highly different structure. In the lower troposphere (850 hPa) over the KMC region, abnormal divergence maintains (Fig. 5c), indicating that abnormal cooling (Figs. 4c, d) in the KMC region can still lead to divergence after the ENSO signal is removed. In the tropical Indian Ocean, the Gill-type response results in a pair of anticyclonic circulation systems roughly symmetric about the equator, which is consistent with the features of abnormal circulation in the lower troposphere during the IOD period. Meanwhile, abnormal anticy-clonic circulation appears in the South China Sea, while abnormal cyclonic circulation appears in the Yangtze River valley and Japan. This structure of circulation anomalies from the South China Sea to Japan is consistent with the EAP/PJ teleconnection pattern (Huang, 1987; Nitta, 1987), which can also be observed at 200 hPa in the upper troposphere (Fig. 5d). Obviously, when the anomalous cyclonic circulation appears in Yangtze River valley, the EASM in that region is weakened.

Furthermore, the divergence still maintains in the KMC region after both the ENSO and IOD signals are removed. As a result, an abnormal wave train structure of circulation appears in East Asia from the South China Sea to Japan (Fig. 5e), while abnormal anticyclonic circulation remains from the Yangtze River valley to Japan. This abnormal circulation pattern also exists at 200 hPa (Fig. 5f), albeit the circulation direction is opposite to that at 850 hPa.

If we set 0.75 as the threshold in the normalized time series INiño3.4 for ENSO, and in IIOD for IOD, then we have 8 neutral years (1979, 1986, 1990, 1993, 1995, 2001, 2003, and 2013) when neither ENSO nor IOD events occurred. It is found that relatively strong convection occurred in the KMC region during June–September 2013, which induced an anomalous circulation pattern (figure omitted) similar to that in Figs. 5e, f, demonstrating the important influences of local convective activity in the KMC region on the EASM.

 Figure 5 Regressions of 850-hPa rotational (streamlines) and divergent (vectors) components of anomalous winds on the time series of (a) Ia, (c) Ib, and (e) Ic. Panels (b), (d), and (f) are the same as (a), (c), and (e), but at 200 hPa. Vectors indicate the areas of anomalous divergent winds at/above the 95% confidence level.

To further clarify the close relations of the OLR variations in the KMC with the EASM, we define an anomalous wind index, Iv850, for the EASM, especially for the wind anomalies over the Yangtze–Huai River valley, by calculating the normalized time series of the anomalous meridional component of 850-hPa winds averaged over the region 25°–35°N, 120°–130°E. This wind index (Iv850) denotes the intensity of the EASM over the Yangtze–Huai River valley. Let Iv850a, Iv850b, and Iv850c be the normalized time series of Iv850 with no filtering, with ENSO removed, and with neither ENSO nor IOD influences, respectively. Then, the correlations among these wind indices and the OLRA averaged over the KMC region are listed in Table 3.

Table 3 Correlations of Iv850a, Iv850b, and Iv850c with Ia, Ib, and Ic. The normalized time series Iv850 is obtained by averaging the meridional component of 850-hPa wind anomalies over the region 25°–35°N, 120°–130°E, which serves as an index of the intensity of the EASM. The absolute value of the correlation coefficient at the 95% confidence level is 0.33
 Index Ia Ib Ic Iv850a –0.52 –0.39 –0.43 Iv850b –0.23 –0.42 –0.47 Iv850c –0.22 –0.41 –0.47

We can see that negative correlation is found between the OLRA in the KMC region and the monsoon intensity over East Asia. This negative correlation is always significant, and does not depend on the removal of the ENSO or IOD signal, or both, from the OLRA time series. This means, even after removing both the ENSO and IOD signals, when convection is stronger than normal, the southerly wind will be stronger over the Yangtze–Huai River area. As seen in Fig. 5, this is in association with the EAP-like wave structure circulation.

5.2 Relationship with East Asian climate variation

Convective activity in the KMC region is closely linked with surface air temperature anomalies through circulation changes. Outputs produced by the regression of air temperature at 2 m above the surface on Ia, Ib, and Ic are shown in Figs. 6a, c, e, respectively. Before the ENSO and IOD signals are removed (Fig. 6a), positive surface air temperature anomalies are found in northwestern MC region, including Sumatra, Malaysia, the South China Sea, and Kalimantan, while negative anomalies appear in other regions of the MC. Apparently, when convection is active in the KMC region, surface temperature is extremely low in southwestern KMC, which is possibly due to the intensified surface evaporation and the blocking of solar radiation by convective clouds. In contrast, surface temperature is abnormally high to the east of 120°E, due to the influence of warm SSTAs. To a certain degree, such higher-than-normal surface temperatures are favorable for the intensification of convection. On the contrary, when convection in the KMC region is weaker than normal, surface air temperature is abnormally warmer over the islands but cooler than normal over the ocean, and the latter is associated with ENSO. Due to the influence of abnormal circulation, negative surface temperature anomalies appear in southwestern China and Japan, which are linked to positive precipitation anomalies in southern China and the islands of Japan (Fig. 6b).

After the ENSO signal is removed, the regression result indicates that there is no significant surface air temperature anomaly in the KMC region, except the area to the south of the equator, where large negative anomalies can be found (Fig. 6c). Negative temperature anomalies in the warm pool region of western Pacific and southeastern China are very weak; positive temperature anomalies are significant in northeastern China, which is related to the large decrease in precipitation in this region (Fig. 6d). Figure 6d shows an abnormally low–high–low precipitation pattern extending from the KMC region northeastward into China, consistent with the abnormal circulation pattern of the EAP/PJ teleconnection shown in Fig. 5c. After the IOD signal is also removed (Figs. 6e, f), no significant changes can be found in temperature and precipitation over the KMC region and its northern side.

 Figure 6 Distributions of regression coefficients of (a, c, e) anomalous air temperature at 2 m above the surface and (b, d, f) anomalous precipitation, as obtained by regression upon the time series of (a, b) Ia, (c, d) Ib, and (e, f) Ic. The stippled areas represent values exceeding the 90% confidence level.

It is clear from the above analysis that abnormally low precipitation always corresponds to an anomalously high OLRA in the KMC region. Meanwhile, negative anomalies in the atmospheric column heating are accompanied by strong descending motion. As a result, anomalous divergence (convergence) occurs in the lower (upper) troposphere, which acts as a source of vorticity and triggers Rossby waves to affect climate variations in East Asia. After the ENSO signal is removed, the EAP/PJ teleconnection pattern in the troposphere induced by abnormal convective activity in the KMC region becomes clearer, and it remains even after the IOD signal is also removed. This indicates that the variation of the OLRA in the KMC region is very important in influencing the EASM circulation anomalies, and even the summer climate conditions, in China and Japan.

6 Summary

Based on the findings of the present study, we can draw the following conclusions:

In boreal summer (June–September), the interannual variability of convection in the MC region is most significant over 10°S–10°N, 95°–145°E, which can be roughly defined as the KMC region. Using regression analysis, we find that large convective activity and precipitation anomalies still exist in this region even after the ENSO and IOD signals are removed. Anomalous convection in the KMC region plays an important role in influencing EASM circulation changes, and thereby affects the climate conditions in summer in East Asian countries.

The correlation coefficient between the area-averaged OLRA and time series of the Niño index (INiño3.4) reaches 0.84, while that with the IOD index (IIOD) is 0.59. Around 70% of the variance of convective activity in the KMC region can be explained by the ENSO signal. Generally, the OLRA in the KMC region presents a 3–7-yr cycle, while ENSO and the IOD mainly affect convective activity in the KMC region with periodic cycles of less than 7 yr. Convective activity in the KMC region also varies with prominent quasi-3-and quasi-7-yr cycles, and such convection bears no close relation with ENSO and the IOD.

When the OLRA is positive in the KMC region, precipitation is abnormally low. Meanwhile, negative anomalies appear in the atmospheric column heating, accompanied by strong descending motion of the air. As a result, low-level divergence and high-level convergence are anomalously induced. Corresponding to these anomalous divergence and convergence, circulation anomalies with a wave-train pattern consistent with the EAP/PJ teleconnection are triggered and propagate northeastward from the KMC region, weakening the EASM, and henceforth affecting the summer climate in East Asia. Moreover, when the OLRA is positive in the KMC region, significant positive precipitation anomalies occur from the Yangtze River valley in China to the islands of Japan. After the ENSO signal is removed, or both the ENSO and IOD signals are removed, the EAP/PJ teleconnection pattern of circulation anomalies becomes much clearer, indicating that convective activity anomalies in the KMC region have significant impacts on the EASM.

It is worth noting that the variance of the OLRA in the KMC region can be largely explained by the Niño3.4 index. However, those that cannot be explained by the ENSO signal seem to be more clearly linked with the EAP/PJ teleconnection pattern. But what is the reason for this phenomenon, and why can only a much smaller part of the OLRA variance over the entire MC region be explained by the Niño3.4 signal? Further studies are needed to address these questions.

The place where the most significant interannual variability of convective activity occurs is found in the MC. Anomalous divergence in association with the OLRA in this area triggers an EAP/PJ-like anomalous wave train that propagates northeastward and leads to significant positive precipitation anomalies from the Yangtze River valley in China to the islands of Japan. This EAP/PJ-like wave pattern becomes even clearer after removal of the ENSO signal and the combined effects of ENSO and IOD, suggesting that convection anomalies over this region have important impacts on the occurrence of EASM anomalies.

The above results are expected to help better understand the mechanisms involved in EASM anomalies, and to provide clues for prediction of summer precipitation anomalies in China.

Acknowledgments. Some data used in this study were provided by the Nanjing Atmospheric Data Center of the Department of Earth Science, Nanjing University of Information Science & Technology. The NCEP–NCAR re-analysis data were from the NOAA's Earth System Research Laboratory (http://www.esrl.noaa.gov/). Figures were plotted by using Grads and NCL.

References
 Alexander A., M. Bladé, I. Newman, et al., 2002: The atmospheric bridge: The influence of ENSO teleconnections on air–sea interaction over the global oceans. J. Climate, 15, 2205–2231. DOI:10.1175/1520-0442(2002)015 < 2205:tabtio > 2.0.co; 2 Araki R., Yamanaka M. D., Murata F., et al., 2006: Seasonal and interannual variations of diurnal cycles of wind and cloud activity observed at Serpong, West Jawa, Indonesia. J. Meteor. Soc. Japan, 84, 171–194. DOI:10.2151/jmsj.84a.171 Ashok Y. Guan, K. Yamagata, 2001: Impact of the Indian Ocean Dipole on the relationship between the Indian monsoon rainfall and ENSO. Geophys. Res. Lett., 28, 4499–4502. DOI:10.1029/2001gl013294 Ashok Y. Guan, K. Yamagata, 2003: A look at the relationship between the ENSO and the Indian Ocean Dipole. J. Meteor. Soc. Japan, 81, 41–56. DOI:10.2151/jmsj.81.41 Baquero-Bernal Latif, A. Legutke, 2002: On dipolelike variability of sea surface temperature in the tropical Indian Ocean. J. Climate, 15, 1358–1368. DOI:10.1175/1520-0442(2002)015 < 1358:odvoss > 2.0.co; 2 Chang C.-P., Wang Z., Ju J. H., et al., 2004: On the relationship between western Maritime Continent monsoon rainfall and ENSO during northern winter. J. Climate, 17, 665–672. DOI:10.1175/1520-0442(2004)017 < 0665:otrbwm > 2.0.co; 2 Chen L., X. J. Zhou, 2014: Relative role of tropical SST forcing in the 1990s periodicity change of the Pacific–Japan pattern interannual variability. J. Geophys. Res., 119, 13043–13066. DOI:10.1002/2014jd022064 Ding Y. H., Li C. Y., He J. H., et al., 2004: South China Sea monsoon experiment (SCSMEX) and the East Asian monsoon. Acta Meteor. Sinica, 62, 561–586. DOI:10.11676/qxxb2004.057 Gill A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447–462. DOI:10.1002/qj.49710644905 Guan Y., Z. Yamagata, 2003: The unusual summer of 1994 in East Asia: IOD teleconnections. Geophys. Res. Lett., 30, 1544. DOI:10.1029/2002gl016831 Hackert C., E. Hastenrath, 1986: Mechanisms of Java rainfall anomalies. Mon. Wea. Rev., 114, 745–757. DOI:10.1175/1520-0493(1986)114 < 0745:MOJRA > 2.0.CO; 2 Hendon H. H., 2003: Indonesian rainfall variability: Impacts of ENSO and local air–sea interaction. J. Climate, 16, 1775–1790. DOI:10.1175/1520-0442(2003)016 < 1775:IRVIOE > 2.0.CO; 2 Huang, R. H. , 1987: Influence of the heat source anomaly over the tropical western Pacific on the subtropical high over East Asia. International Conference on the General Circulation of East Asia, Chengdu, China, 10-15 April, 40-51. Huang R. H., 1990: Studies on the teleconnections of the general circulation anomalies of East Asia causing the summer drought and flood in China and their physical mechanism. Chinese J. Atmos. Sci., 14, 108–117. DOI:10.3878/j.issn.1006-9895.1990.01.14 Huo W., L. C. Jin, 2016: The interannual relationship between anomalous precipitation over southern China and the south eastern tropical Indian Ocean sea surface temperature anomalies during boreal summer. Atmos. Sci. Lett., 17, 610–615. DOI:10.1002/asl.710 Jin C., D. N. Hameed, S. W. Huo, 2016: Recent changes in ENSO teleconnection over the western Pacific impacts the eastern China precipitation dipole. J. Climate, 29, 7587–7598. DOI:10.1175/JCLI-D-16-0235.1 Kalnay E., Kanamitsu M., Kistler R., et al., 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc., 77, 437–471. DOI:10.1175/1520-0477(1996)077 < 0437:tnyrp > 2.0.co; 2 Kanamitsu M., Ebisuzaki W., Woollen J., et al., 2002: NCEP–DOE AMIP-Ⅱ reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 1631–1643. DOI:10.1175/BAMS-83-11-1631 Kao Yu, 2009: Contrasting eastern-Pacific and central-Pacific types of ENSO. J. Climate, 22, 615–632. DOI:10.1175/2008jcli2309.1 Li T., 2014: Recent advance in understanding the dynamics of the Madden–Julian oscillation. J. Meteor. Res., 28, 1–33. DOI:10.1007/s13351-014-3087-6 Liu H., C. W. Moncrieff, 1998: A numerical study of the diurnal cycle of tropical oceanic convection. J. Atmos. Sci., 55, 2329–2344. DOI:10.1175/1520-0469(1998)055 < 2329:ansotd > 2.0.co; 2 Luo B., H. Yanai, 1984: The large-scale circulation and heat sources over the Tibetan Plateau and surrounding areas during the early summer of 1979. Part Ⅱ: Heat and moisture budgets. Mon. Wea. Rev., 112, 966–989. DOI:10.1175/1520-0493(1984)112 < 0966:tlscah > 2.0.co; 2 Madden A., R. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50-day period. J. Atmos. Sci., 29, 1109–1123. DOI:10.1175/1520-0469(1972)029 < 1109:dogscc > 2.0.co; 2 McBride L., J. R. Haylock, M. Nicholls, 2003: Relationships between the Maritime Continent heat source and the El Niño–Southern Oscillation phenomenon. J. Climate, 16, 2905–2914. DOI:10.1175/1520-0442(2003)016 < 2905:rbtmch > 2.0.co; 2 Meng X. Wu, 2000: Gearing between the Indo-monsoon circulation and the Pacific–Walker circulation and the ENSO. Part Ⅱ: Numerical simulation. Chinese J. Atmos. Sci, 24, 15–25. Mori I. Hamada, S. I. Tauhid, et al., 2004: Diurnal land–sea rainfall peak migration over Sumatra island, Indonesian Maritime Continent observed by TRMM satellite and intensive rawinsonde soundings. Mon. Wea. Rev., 132, 2021–2039. DOI:10.1175/1520-0493(2004)132 < 2021:dlrpmo > 2.0.co; 2 Neale Slingo, 2003: The Maritime Continent and its role in the global climate: A GCM study. J. Climate, 16, 834–848. DOI:10.1175/1520-0442(2003)016 < 0834:tmcair > 2.0.co; 2 Nitta T., 1987: Convective activities in the tropical western Pacific and their impact on the Northern Hemisphere summer circulation. J. Meteor. Soc. Japan, 65, 373–390. DOI:10.2151/jmsj1965.65.3_373 Oh Kim, J.-H. Lim, 2012: Impact of MJO on the diurnal cycle of rainfall over the western Maritime Continent in the austral summer. Climate Dyn., 38, 1167–1180. DOI:10.1007/s00382-011-1237-4 Oh J.-H., Kim B.-M., Kim K.-Y., et al., 2013: The impact of the diurnal cycle on the MJO over the Maritime Continent: A modeling study assimilating TRMM rain rate into global analysis. Climate Dyn., 40, 893–911. DOI:10.1007/s00382-012-1419-8 Qian W. H., Hu H. R., Deng Y., et al., 2012: Signals of interannual and interdecadal variability of air–sea interaction in the basin-wide Indian Ocean. Atmosphere-Ocean, 40, 293–311. DOI:10.3137/ao.400302 Ramage C. S., 1968: Role of a tropical " Maritime Continent" in the atmospheric circulation. Mon. Wea. Rev., 96, 365–370. DOI:10.1175/1520-0493(1968)096 < 0365:roatmc > 2.0.co; 2 Saji H., N. Yamagata, 2003: Possible impacts of Indian Ocean Dipole mode events on global climate. Climate Research, 25, 151–169. DOI:10.3354/cr025151 Saji N. H., Goswami B. N., Vinayachran P. N., et al., 1999: A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363. DOI:10.1038/43854 Shibagaki Y., Shimomai T., Kozu T., et al., 2006: Multiscale aspects of convective systems associated with an intraseasonal oscillation over the Indonesian Maritime Continent. Mon. Wea. Rev., 134, 1682–1696. DOI:10.1175/mwr3152.1 Song L. Gordon, 2004: Significance of the vertical profile of the Indonesian throughflow transport to the Indian Ocean. Geophys. Res. Lett., 31, L16307. DOI:10.1029/2004gl020360 Tao, S. Y. , and L. X. Chen, 1987: A review of recent research on the East Asian summer monsoon in China. Monsoon Meteorology, C. -P. Chang, T. N. Krishnamurti, Eds. Oxford, Oxford University Press, 60-92. Teo C.-K., Koh T.-Y., Lo J. C.-F., et al., 2011: Principal component analysis of observed and modeled diurnal rainfall in the Maritime Continent. J. Climate, 24, 4662–4675. DOI:10.1175/2011jcli4047.1 Torrence P. Compo, 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79, 61–78. DOI:10.1175/1520-0477(1998)079 < 0061:apgtwa > 2.0.co; 2 Torrence J. Webster, 1999: Interdecadal changes in the ENSO-monsoon system. J. Climate, 12, 2679–2690. DOI:10.1175/1520-0442(1999)012 < 2679:icitem > 2.0.co; 2 Wang Lee, B. Q. Xiang, 2015: Asian summer monsoon rainfall predictability: A predictable mode analysis. Climate Dyn., 44, 61–74. DOI:10.1007/s00382-014-2218-1 Wang J., L. H. Guo, S. Ge, 2016: The timing of South Asian high establishment and its relation to tropical Asian summer monsoon and precipitation over east–central China in summer. J. Trop. Meteor., 22, 136–144. Wu Hsu, 2009: Topographic influence on the MJO in the Maritime Continent. J. Climate, 22, 5433–5448. DOI:10.1175/2009jcli2825.1 Wu J., Xu X. F., Jin F. F., et al., 2013: Research of the intraseasonal evolution of the East Asian Pacific pattern and the maintenance mechanism. Acta Meteor. Sinica, 71, 476–491. DOI:10.11676/qxxb2013.038 Xie P., P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558. DOI:10.1175/1520-0477(1997)078 < 2539:gpayma > 2.0.co; 2 Yanai Esbensen, M. Chu, 1973: Determination of bulk properties of tropical cloud clusters from large-scale heat and moisture budgets. J. Atmos. Sci., 30, 611–627. DOI:10.1175/1520-0469(1973)030 < 0611:dobpot > 2.0.co; 2 Yang L. Jia, H. Y. Li, 2006: The tropical Pacific–Indian Ocean temperature anomaly mode and its effect. Chinese Sci. Bull., 51, 2878–2884. DOI:10.1007/s11434-006-2199-5 Zhou Murtugudde, 2010: Influences of Madden–Julian oscillations on the eastern Indian Ocean and the maritime continent. Dyn. Atmos. Oceans, 50, 257–274. DOI:10.1016/j.dynatmoce.2009.12.003