Abstract

The influence of the MJO on the phase and amplitude of the diurnal cycle of rainfall during Australian summer [December–February (DJF)] over the Maritime Continent (MC) and northern Australia is investigated using the Tropical Rainfall Measuring Mission (TRMM) 3B42 and 3G68 datasets. The gridded rainfall was partitioned into MJO categories (active, suppressed, and weak) based on their longitudinal position and by utilizing the real-time multivariate MJO (RMM) index of Wheeler and Hendon. The diurnal cycles were composited and an empirical orthogonal function (EOF) analysis was applied to extract the spatial and temporal variability.

Distinct variations in the rainfall distribution pattern among categories of the MJO over land and ocean are seen. The result of the composite-mean rainfall distribution shows that the average daily rainfall rate over islands is higher during suppressed MJO days, while for surrounding oceans and northern regions of Australia, more rainfall occurs during MJO active days. The normalized relative amplitude (NRA) of the diurnal cycle of rainfall shows that morning rainfall near coastal areas during active days of the MJO is 1.5 times greater than the climatological-mean rainfall but is less than or equal to the climatological mean during other phases of the MJO. Similarly, during the suppressed phase of the MJO evening rainfall is greater over the islands than in other MJO phases. The first two modes of the EOF alone explain more than 88% (65%) of the variance for the 3B42 (3G68) rainfall, and the corresponding principal component time series show a marked diurnal cycle. The results show that both the amplitude and phase of the diurnal cycle of rainfall are modulated by the categories of the MJO. In general, the peak in the diurnal cycle for active (suppressed/weak) days of the MJO lags (leads) the peak in the diurnal cycle for total rainfall by 2 h. Over Darwin and its adjacent regions, the active phase of the MJO is responsible for the occurrence of maximum rainfall after midnight, which is unusual in this region.

1. Introduction

The study of the diurnal cycle of rainfall and convection has been the subject of fundamental research over several decades, especially in the tropics because of their important role in the global and regional transport of water vapor and energy (Slingo et al. 2004; Sakurai et al. 2005). A significant amount of uptake and release of latent heat is associated with the diurnal cycle of tropical rainfall (Mori et al. 2004), which then affects synoptic and mesoscale weather phenomena. Therefore, understanding the mechanisms of the diurnal cycle of rainfall in the tropics is important in the study of variations in large-scale circulations such as the Asian and Australian monsoons.

In the tropics in general, the diurnal cycle of rainfall and convection tends to peak during the evening over land but during the morning over oceans. However, this general rule is not applicable everywhere, and previous results have shown several interesting regional variations, with maximum rainfall in some locations occurring during morning (afternoon) hours over land (ocean). Many hypotheses have been proposed so far to explain the diurnal maxima in rainfall and convection, as documented in Yang and Smith (2006). The prime mechanism responsible for the continental surface rainfall maximum during mid- to late afternoon is the diurnally regulated surface solar radiative heating, which has two major effects on the atmosphere. The first is the static destabilization of the atmosphere due to sensible and latent heating from the surface. The second is the formation of mesoscale circulations due to the differential heating caused by horizontal variations in land surface heating. The propagation of rainfall away from its afternoon origin over high topography leads to late evening–early morning (LE–EM) maxima in rainfall over large landmasses, termed by Yang and Smith (2006) the mobile terrain–forced rainfall system mechanism [see also Tripoli and Cotton (1989a,b)]. In addition, over land there is enhanced nighttime radiative cooling (NRC) and a consequent increase in relative humidity (Dai 2001). Together these are known as the continental-based NRC–elevated relative humidity (ERH) mechanism. Despite these hypotheses, there is no widely accepted explanation for this LE–EM maximum in rainfall over continents. Similarly, over oceans, there are a number of mechanisms that have been proposed to explain the endemic early morning rainfall maximum. For example, the static radiation convection mechanism assumes that the morning maximum rainfall is due to enhanced cloud-top IR cooling at night and a consequent increase in the thermal lapse rate, provided that clouds already exist (e.g., Kraus 1963). In contrast, the dynamic radiation convection mechanism (Gray and Jacobson 1977) is based on the large daytime–nighttime differences in the radiative cooling profiles of mesoscale cloud regions in comparison with their surrounding clear-sky regions, resulting in daytime suppression of convection and more rain during night. Similarly, the NRC–ERH mechanism over the ocean is also considered to be one possible explanation for the well-established early morning oceanic surface rainfall maximum. In contrast, under clear-sky conditions over the ocean, a mid- to late afternoon oceanic rainfall maximum is sometimes observed. The ocean surface heating mechanism, which is related to diurnal variations in sea surface temperature (SST), is proposed as a possible cause (Gray and Jacobson 1977).

Despite our understanding of the above mechanisms, many of the current numerical weather prediction (NWP) and climate models still have problems in correctly simulating the phase and amplitude of diurnal variations (Shin et al. 2007). In addition, the characteristics of the diurnal cycle differ from location to location, and in any event, they are modified by dynamics, local orography, and the initiation, propagation, and decay of mesoscale convective systems. The mechanisms so far proposed are exclusively based on the analysis of the diurnal cycle composited for the total rainfall and much less attention has been paid to the scale interaction of rainfall (Slingo et al. 2003) and its modulation by intraseasonal- (Ichikawa and Yasunari 2008) and interannual-scale phenomena (Yang et al. 2009). According to Slingo et al. (2003), scale interaction is defined as the process that describes the influence of the large-scale and low-frequency variability on small spatial scale high-frequency variability of the climate system and vice versa. The Madden–Julian Oscillation (MJO) and El Niño–Southern Oscillation (ENSO) are considered to be large spatial scale low-frequency variability, whereas the diurnal cycle is a small spatial scale high-frequency atmospheric phenomenon. In this study, we examine the modulations of the diurnal cycle of rainfall by the MJO.

First noted in the early 1970s (Madden and Julian 1971, 1994), the MJO is a remarkable feature of the atmospheric circulation and moist convection that involves planetary-scale regions of enhanced and suppressed rainfall. Madden and Julian (1994) describe the MJO as the most dominant intraseasonal climate variation in the tropics. It exhibits a distinctive multiscale structure, geographic preferences, seasonal cycle, and interannual variability. The combination of these primary features distinguishes the MJO from other types of intraseasonal phenomena in the tropics. It propagates eastward at an average speed of 5–10 m s−1 across the equatorial Indian and western-central Pacific Oceans with a local intraseasonal period of 30–90 days (Zhang 2005). Many of the current generation general circulation models (GCMs) have not been able to produce the correct amplitude and phases of the MJO, with some exceptions. For example, Rashid et al. (2010) found that the MJO can be predicted using the Predictive Ocean–Atmosphere Model for Australia (POAMA) out to 2–3 weeks, with amplitude slightly above and propagation speed slightly slower than observed. The major difficulty in simulating the MJO with GCMs appears to be inadequacies in the cumulus parameterization used to estimate the vertical redistribution of heat and moisture by unresolved convective clouds in GCMs (Slingo et al. 1996; Lin et al. 2006; Miura et al. 2007).

A number of studies (e.g., Chen and Houze 1997; Yang and Slingo 2001; Slingo et al. 2003; Tian et al. 2004, 2006; Sakurai et al. 2005; Ichikawa and Yasunari 2006, 2008) have investigated scale interaction between the diurnal cycle and the MJO. Chen and Houze (1997) showed that the diurnal cycle of tropical deep convective cloud systems during the convectively active (suppressed) phase of the MJO over the open oceans is slightly larger (smaller) with an early morning (afternoon) maximum, although the measured magnitude of the diurnal cycle is a strong function of the chosen IR temperature threshold chosen Slingo et al. (2003) indicated that the diurnal cycle in SST during the suppressed phase of the MJO leads to a triggering of cumulus congestus clouds, which serve to moisten the free troposphere and hence precondition the atmosphere for the next active phase. Tian et al. (2006) found that the diurnal cycle of tropical convective cloud amount (DCC) is enhanced (reduced) over both land and ocean during the convectively active (suppressed) regime of the MJO. However, they noticed no significant differences in the phase of DCC for active and suppressed MJO regimes. Ichikawa and Yasunari (2008) observed that the diurnal cycle of rainfall over and around New Guinea shows a systematic modulation associated with intraseasonal variability in the large-scale circulation pattern, with differences in the phase of the diurnal cycle for easterly (suppressed) and westerly (active) wind regimes. They also found that the speed and extent of propagation of rainfall differ depending on the phase of the intraseasonal oscillation. The above studies have advanced our understanding of the impact of the MJO on the diurnal cycle of rainfall and tropical convection. Nevertheless, these studies were mainly based on a few MJO events, which can differ significantly from event to event. In addition, metrics used to partition the MJO phases were not consistent from study to study. These issues necessitate reexamining the climatology of the MJO impact on the diurnal cycle using a high space and time resolution rainfall dataset.

In this study, a region (Fig. 1a) between 10°N and 30°S and 95° and 160°E is defined to diagnose the modulation by the MJO of the amplitude and phase of the diurnal cycle of rainfall over land, coastal, and ocean regions. The area includes the Maritime Continent (MC) and northern Australia, where not only is the diurnal cycle of rainfall prominent (Ohsawa et al. 2001; Mori et al. 2004; Kikuchi and Wang 2008) but substantial intraseasonal and interannual variation also occurs (Lau and Waliser 2005; Ichikawa and Yasunari 2006, 2008; Wheeler et al. 2009) during the Australian summer season [December–February (DJF)]. The diurnal range, which is defined by Kikuchi and Wang (2008) as the difference between climatological daily maximum and climatological daily minimum rainfall, is also very large over the MC. There have been a number of studies on the diurnal cycle of total rainfall over individual islands of the MC utilizing satellite-derived rainfall and convection (Mori et al. 2004; Sakurai et al. 2005) and model-simulated rainfall (Zhou and Wang 2006; Qian 2008; Wu et al. 2009), but very few have examined the influence of the intraseasonal oscillation on the amplitude and phase of the diurnal cycle of rainfall (Ichikawa and Yasunari 2006, 2008).

Fig. 1.

(a) Location map of study area showing topography at 500-m contour intervals over land (white outline) and bathymetrical features in shading and (b) the climatological-average 3B42 rainfall (mm day−1) composited from 10 (1998–2008) DJF seasons. Also shown in Fig. 1a are locations of cross sections for the analysis of diurnal rainfall propagation anomalies shown in Figs. 13 and 15.

Fig. 1.

(a) Location map of study area showing topography at 500-m contour intervals over land (white outline) and bathymetrical features in shading and (b) the climatological-average 3B42 rainfall (mm day−1) composited from 10 (1998–2008) DJF seasons. Also shown in Fig. 1a are locations of cross sections for the analysis of diurnal rainfall propagation anomalies shown in Figs. 13 and 15.

The performance of most models in simulating these activities is also very poor in the MC (Neale and Slingo 2003), and understanding the influence of the MJO on the diurnal cycle and vice versa will lead to improvements in model simulation. In particular, many models do not tend to capture the passage of the MJO through the MC and produce an MJO that is too weak and moves too quickly (Hartmann and Hendon 2007). Miura et al. (2007) found that the Nonhydrostatic Icosahedral Atmospheric Model (NICAM; Satoh et al. 2008) is able to simulate a realistic representation of the MJO using a 3.5-km- and a 7-km-grid global cloud-resolving version. For the same model configuration, Sato et al. (2009) find that a 14-km-grid NICAM run simulates the maxima in the diurnal cycle of rainfall, but this is delayed by several hours compared with observations. However, a 3.5-km-grid NICAM run simulates the phase of the diurnal cycle of rainfall fairly well but slightly overestimates the amplitude, though they only run the model with a 3.5-km grid for 7 days from 25 December 2006, which was an MJO active period (as defined in section 2). In contrast, the diurnal cycle of rainfall over the major islands (horizontal scale greater than 200 km) of the MC simulated by a 20-km-grid version of the Meteorological Research Institute (MRI) GCM indicates that the maximum simulated rainfall occurred in the early afternoon on these islands, while the observed rainfall has its maximum at night (Hara et al. 2009).

The objectives of the present study are focused on increasing our present understanding of the causes of variation in the diurnal cycle in the above regions by inferring the different characteristics of the diurnal cycle of rainfall during the MJO phases. This study examines this issue over a wider region than previous studies, using an improved dataset. We also intend to provide a foundation for the validation of future results from regional-scale numerical models and high-resolution climate or cloud-resolving models in the above region. This study utilizes high spatial and temporal resolution products from the Tropical Rainfall Measuring Mission (TRMM) satellite for 10 Southern Hemispheric summer seasons (DJF) and the real-time multivariate MJO (RMM) index from Wheeler and Hendon (2004), along with other data and methodology as described in section 2. The climatological-average pattern and the pattern associated with the differing regimes of the MJO for rainfall and other atmospheric variables are shown in section 3. The influence of the MJO on the diurnal cycle of rainfall is explained in section 4. Section 5 summarizes the major findings of this study, discusses possible explanations for the results, and suggests future work.

2. Data and methods

a. TRMM satellite data

The present study utilizes two TRMM rainfall products, namely 3G68_V6 and 3B42_V6 (Huffman et al. 2007), for 10 full Australian summer seasons (1998–2008). The details of the TRMM sensors and algorithms are available in Kummerow et al. (1998). The TRMM 3G68_V6 dataset was obtained online from ftp://trmmopen.gsfc.nasa.gov/pub/. The 3G68 dataset is an hourly gridded text [American Standard Code for Information Interchange (ASCII)] product containing TRMM instrumental rain estimates at 0.50° × 0.50° horizontal resolution. It consists of total pixels, rainy pixels, mean rain rate (mm hr−1), and the percentage of rainfall calculated to be convective from the 2A12 [TRMM Microwave Imager (TMI)], 2A25 [precipitation radar; (PR)], and 2B31 (TMI–PR combined) algorithms merged into a single daily file. We analyzed both the PR and TMI datasets, but only the results from the PR sensor are discussed in this study, as the TMI data have a systematic error over coastal areas in which both land and ocean regions are included in a single pixel (Mori et al. 2004). Although the PR sensor covers only a small region at a time and has coarse sampling time intervals because of its narrower swath, it does provide a direct satellite estimate of rainfall in this region. The 3B42_V6 dataset was obtained from the TRMM Science Data and Information System (TSDIS), distributed by the National Aeronautics and Space Administration (NASA) Goddard Distributed Active Archive Center (DAAC). A description of the 3B42 algorithm is provided online at http://trmm.gsfc.nasa.gov/3b42.html. The 3B42 data contains the estimated rain rate (mm hr−1) created by calibrating the IR brightness temperatures to the high-quality microwave estimates. The 3B42 rainfall product has 3-h temporal resolution with 0.25° × 0.25° spatial resolution globally, extending from 50°S to 50°N latitude and from 1998 to present.

b. RMM index and discrimination of MJO phases

The RMM index developed by Wheeler and Hendon (2004) has been used to discriminate the MJO phases. The index is based on a pair of empirical orthogonal functions (EOFs) of the combined fields of near-equatorially 15°S–15°N averaged 850-hPa zonal wind, 200-hPa zonal wind, and outgoing longwave radiation (OLR) data (called RMM1 and RMM2). When the MJO is in a strong cycle (RMM amplitude greater than 1), deep convection is shown to propagate eastward from the Indian Ocean (phases 2–3) to the MC (phases 4–5) and over the western Pacific (phases 6–7), before decaying around the date line in the central Pacific (phases 8 and 1; Wheeler and Hendon 2004). The TRMM data that we analyzed displayed a clear difference in the location of higher rainfall intensities during all phases of the MJO (not shown). Hidayat and Kizu (2010) have also noted similar coherent eastward propagation of rainfall anomalies during the passage of the MJO over the MC.

Over the study period in our analysis, the total number of days falling into each of the MJO phases 1–8 is 26, 65, 156, 128, 85, 85, 70, and 43 days, respectively, with the weak phase of the MJO, defined as RMM amplitude less than 1, consisting of 245 days. Since there are not enough observations of rainfall (especially in 3G68) to composite the diurnal cycle for the eight individual phases of the MJO, the hourly–three-hourly rainfall data are instead classified into just three categories based on (i) the occurrence of an MJO event (i.e., a strong or weak MJO cycle at a given time) and (ii) the distribution of MJO rainfall partitioned by the phase diagram of Wheeler and Hendon (2004) during a strong MJO cycle (i.e., if active or suppressed convection is occurring at a given location). Hence, hereafter the MJO at a given time and location is described as being active, suppressed, or weak as shown in Table 1. For example, at Darwin and nearby locations during a strong MJO cycle, active convection occurs when the MJO is in phases 4, 5, and 6, and suppressed convection occurs when the MJO is in phases 7–8 and 1–3. At all other times the MJO is defined to be in a weak cycle. A composite dataset for the diurnal cycle of rainfall is made at each grid point (local time) for both datasets. A 4-h running mean (Negri et al. 2002) is applied to the diurnal cycle composited from the 3G68 data to further smooth the data. An EOF analysis (Lorenz 1956; Hannachi et al. 2007) is applied to the composited diurnal cycle of rainfall to identify any patterns in the data. The significance of each eigenvalue is further evaluated according to North’s rule of thumb (North et al. 1982) and shown by vertical error bars.

Table 1.

MJO active and suppressed categories by longitude for the MJO phase calculated using the method of Wheeler and Hendon (2004).

MJO active and suppressed categories by longitude for the MJO phase calculated using the method of Wheeler and Hendon (2004).
MJO active and suppressed categories by longitude for the MJO phase calculated using the method of Wheeler and Hendon (2004).

c. NCEP reanalysis II data

This study also uses the National Centers for Environmental Prediction (NCEP)–Department of Energy (DOE) reanalysis II dataset (Kanamitsu et al. 2002) to explain the links between the large-scale circulation and the rainfall variations. Variables examined were air temperature, wind (U, V), vertical velocity (omega), and specific and relative humidity. In addition, the National Oceanic and Atmospheric Administration (NOAA) interpolated OLR data (Liebmann and Smith 1996) have been used to identify the location of deep convection. All these data are daily average values at pressure levels (except for OLR) with a spatial resolution of 2.5° × 2.5° and were acquired from NOAA/Office of Atmospheric Research (OAR)/Earth System Research Laboratory (ESRL) Physical Sciences Division (PSD), Boulder, Colorado (available online from http://www.esrl.noaa.gov/psd/). These parameters were also composited according to the three MJO categories to produce the average pattern for each category, as discussed in section 2b.

3. General features of rainfall and atmospheric circulation

a. Climatological-average spatial pattern

Figure 1b shows the climatological average for 10 DJF seasons of the total rainfall rate (mm day−1) composited from the 3B42 data over the study area. On comparing the total rainfall with the topographical features of the study area (Fig. 1a), orographic effects are evident in the spatial distribution of the rainfall (Fig. 1b). On average, the islands of the MC and northern coastal regions of Australia receive more rainfall (Fig. 1b) in comparison to the shallow and open sea regions. However, there are some recognized deficiencies in the geographical pattern of these data. For instance, in northeastern Australia, ground observations imply that maximum rainfall during the DJF season occurs near 15°S, 145°E, whereas Fig. 1b shows maximum values near 13°S, 143°E, although a relative maximum is also observed in this location in station data. Localized maxima in rainfall appear over the sea in the vicinity of southwestern Sumatra, over the sea northeast of New Guinea, and over the eastern Java Sea between Java, Sulawesi, and Borneo. Wu et al. (2009) found that abundant rainfall over the sea west of Sumatra Island is due to the mountains on the island and the resultant thermally and convectively induced local circulations. The distribution of rainfall over New Guinea is also controlled by the presence of high mountains (>3000 m) that extend northwest to southeast, depositing more rainfall over the southern plain and along mountain ridges. A similar effect on the distribution of rainfall due to orography and the large-scale wind over the islands has been documented over Sumatra (Mori et al. 2004), Java (Qian 2008), Borneo (Ichikawa and Yasunari 2006; Wu et al. 2008), and New Guinea (Zhou and Wang 2006; Ichikawa and Yasunari 2008). The results are consistent in the 3G68 product (not shown) except that the spatial pattern is noisier than 3B42, which is expected because of a smaller number of observations.

Inhomogeneity in the spatial distributions of the mean atmospheric conditions can be seen in Fig. 2, even at the coarse grid resolution of these data. Figure 2a shows the mean OLR (as a proxy for deep convection) and average winds at 700 hPa. Low OLR values (<220 W m−2) are more closely associated with deep atmospheric convection that corresponds to heavier rainfall events in the tropics, even though there is some contribution to these low OLR values from tropical cirrus (e.g., Massie et al. 2010). Figure 2b shows the mean moisture convergence and average wind at 850 hPa. The deep convection and associated moisture convergence mainly occur in southern latitudes in a band from the Indian Ocean extending toward the Pacific Ocean. A clear intrusion of moisture from the northern Australia into the inland areas in an inverted V shape can be seen in Fig. 2b. The strong moisture convergence south of the Kimberley Plateau (20°S, 130°E) may partially be related to the adjacent heat low (indicated by the wind field in Fig. 2c). In contrast, the main areas of deep convection appear only over the northernmost land area of northern Australia (Fig. 2a). The loci of low OLR values and strong moisture convergence are coincident with the general region of large rainfall (cf. Fig. 1b). Figure 2c shows the vertical velocity at 500 hPa and average wind at 925 hPa. The vertical velocity is strongest (>0.08 Pa s−1) and confined in a band at 5°N–5°S until it reaches the western tip of New Guinea (125°E), while east of this it widens and also covers northern Australia. However, the strongest ascent (>0.12 Pa s−1) is collocated with the regions of high rainfall such as over Sulawesi and over the sea northeast of New Guinea.

Fig. 2.

Climatological-mean features of the study area for (a) OLR (W m−2) and 700-hPa wind (m s−1), (b) moisture convergence (q · V; ×10−8 s−1) and wind at 850 hPa, and (c) 500-hPa omega (Pa s−1) and wind at 925 hPa derived from NCEP–DOE reanalysis II data for 10 (1998–2008) DJF seasons.

Fig. 2.

Climatological-mean features of the study area for (a) OLR (W m−2) and 700-hPa wind (m s−1), (b) moisture convergence (q · V; ×10−8 s−1) and wind at 850 hPa, and (c) 500-hPa omega (Pa s−1) and wind at 925 hPa derived from NCEP–DOE reanalysis II data for 10 (1998–2008) DJF seasons.

The typical pattern of winds at 925 hPa (Fig. 2c) shows northeasterly winds over the Northern Hemisphere that turn counterclockwise across the equator during the austral summer season due to the change in the sign of the Coriolis parameter but also (locally) due to blocking and deflection by the terrain of the Malay Peninsula–Sumatra (Chang et al. 2005). On average, westerly winds dominate just south of the equator over the islands of the MC. The westerly winds become stronger as they pass eastward along the MC, advecting moisture from the Indian Ocean and converging east of New Guinea with easterly winds coming from the Pacific Ocean. The low-level westerly wind forms a cyclonic circulation around the northern part of the Australian continent and is associated with the Australian monsoon trough (Ichikawa and Yasunari 2008). Weaker 925-hPa winds are centered over the sea south of Sumatra, but Fig. 2a shows weaker winds at 700 hPa are located over the sea north and northeast of New Guinea. The strength of the 925-hPa northeasterly wind decreases at higher levels (850 and 700 hPa) and turns to an easterly wind in the Northern Hemisphere, whereas the surface westerly wind strengthens across the southern latitudes of the MC and at 700 hPa extends toward the equator.

b. Spatial patterns for MJO categories

The general features of rainfall are strongly modified during the three MJO categories. Figure 3 depicts the regional distributions of mean rainfall anomaly (3B42) for the three MJO categories (active, suppressed, and weak). During the active MJO (Fig. 3a), positive rainfall anomalies cover broad areas of the study domain, except for the islands of the MC. More specifically, large positive rainfall anomalies (>4 mm day−1) are located over the sea off the southwestern coast of Sumatra, as well as over the South China Sea and the Java, Timor, and Arafura Seas. Similarly, large rainfall anomalies are also concentrated over the sea northeast and east of New Guinea, in the region between New Guinea and northern Australia, and in the Gulf of Carpentaria. Positive rainfall anomalies are also observed over the Malay Peninsula and the Australian continent. The mountains of western Sumatra play an important role in trapping moisture from the Indian Ocean during westerly wind (active MJO) regimes, creating greater rainfall amounts in western districts of the MC. The model results of Wu et al. (2009) show that abundant rainfall over the sea west of Sumatra Island is due to the convergence of thermally and convectively induced local circulations with the low-level synoptic-scale westerly winds. However, the impact of the MJO on the spatial distribution of rainfall is not clear from their results. In contrast, negative rainfall anomalies are found over the southern plain of Kalimantan (Borneo), New Guinea, and over Sulawesi.

Fig. 3.

Rainfall anomalies (mm day−1) for the three MJO categories: (a) active, (b) suppressed, and (c) weak, using 3B42 data.

Fig. 3.

Rainfall anomalies (mm day−1) for the three MJO categories: (a) active, (b) suppressed, and (c) weak, using 3B42 data.

The opposite is seen during suppressed days of the MJO (Fig. 3b), with positive rainfall anomalies located mainly over the islands of the MC, though the amplitude of these anomalies is not as high as it is over the sea in the active MJO. During suppressed days, surface heating causes strong convection over land, but this appears to be amplified in certain regions by convergence zones (Figs. 5b and 6b) causing high intensity rainfall (>12 mm day−1; not shown) to be mainly concentrated over the large islands of the MC (Fig. 3b). The enhanced rainfall during suppressed days over Sulawesi is due to strong moisture transport from the northeast and the presence of orography. During the MJO weak cycle (Fig. 3c), a wider distribution of positive rainfall anomalies than in suppressed days can be observed, which are mostly concentrated near coastal regions and also over the islands of the MC. For convective rainfall, the spatial distribution pattern by MJO categories is similar to that of Fig. 3 (not shown).

Fig. 5.

As in Fig. 4, but for 850-hPa moisture convergence and winds.

Fig. 5.

As in Fig. 4, but for 850-hPa moisture convergence and winds.

Fig. 6.

As in Fig. 4, but for 500-hPa omega and 925-hPa winds.

Fig. 6.

As in Fig. 4, but for 500-hPa omega and 925-hPa winds.

Distinct variations in the atmospheric and thermodynamic conditions are evident in the climatological-mean state for the three MJO categories. The left panel of Fig. 4 shows the mean values of OLR for the three MJO categories, while the right panel gives OLR anomalies. During active days, OLR values less than 220 W m−2 cover a wide area between 8°N and 16°S, accompanied by strong westerly winds at 700 hPa (Fig. 4a). The large negative anomaly (<15 W m−2) in OLR in this location (Fig. 4b) is collocated with regions of positive rainfall anomaly. The suppressed and weak MJO regimes (Figs. 4d and 4f) are associated with widespread positive OLR anomalies consistent with suppressed convection. OLR values less than 220 W m−2 (Fig. 4c) reveal that convection associated with very weak winds is located over Sumatra, Borneo, and New Guinea, which are regions that have more rainfall during suppressed days than active days. During weak days (Fig. 4e), the deep convection has a similar pattern as for suppressed days but with a stronger westerly wind over the MC and smaller OLR values over some of the larger MC islands. For the anomalies, Figs. 4b and 4d show the predominance of the strong westerly (easterly) wind anomalies at 750 hPa between the equator and 12°S during active (suppressed) days, while the winds in the weak case (Fig. 4f) are very close to the climatology (Fig. 2a).

Fig. 4.

(left) Mean OLR and 700-hPa mean winds. (right) Corresponding anomalies from the climatological average for 10 DJF for the three MJO categories: (a),(b) active, (c),(d) suppressed, and (e),(f) weak.

Fig. 4.

(left) Mean OLR and 700-hPa mean winds. (right) Corresponding anomalies from the climatological average for 10 DJF for the three MJO categories: (a),(b) active, (c),(d) suppressed, and (e),(f) weak.

The mean and anomaly patterns of large-scale moisture convergence and wind at 850 hPa for different MJO regimes are shown in Fig. 5. They show that the strength of moisture convergence (right column of Fig. 5) is stronger (weaker) over the major islands of the MC during the suppressed (active/weak) regime of the MJO. Locations of maximum moisture convergence (Fig. 5) are coincident with locations of higher rainfall, as shown in Fig. 3. During the active regime, the moisture convergence over the western coast of Sumatra Island in the active regime is due to the turning of northeasterly wind across the equator merging with the prevailing westerlies, as shown in Fig. 5a. There is also a band of moisture convergence extending southeast from the eastern coast of Kalimantan toward northern Australia. A tongue of moisture also converges poleward from northern Australia, indicating the occurrence of the Australian monsoon during active days, which is more apparent in the anomaly plot of Fig. 5b. Strong westerly winds extend all the way to the eastern edge of the domain and dominate the flow as far south as the “top end” of Australia. The circulation turns clockwise over the Gulf of Carpentaria and weakens as it passes east of New Guinea, resulting in localized strong convergence over the top end of Australia (Fig. 5b). During suppressed days, strong easterly and northeasterly winds dominate over the Northern Hemisphere regions of the domain (Fig. 5c). The northeasterly wind crosses the Malay Peninsula and is deflected eastward, producing more rainfall on the windward side of Sumatra. The easterlies penetrate almost to the center of New Guinea, merging with the westerlies in a convergence zone, resulting in more rainfall toward the south of the New Guinean mountains. The maximum anomaly can be seen over the central part of New Guinea (Fig. 5d). Figures 5e and 5f show that the weak regime displays very close to the climatological average of moisture convergence.

Substantial midlevel (500 hPa) ascent (omega, Pa s−1) covers a wide region of the study area during the active MJO regime (Fig. 6a). A notable vertical velocity anomaly (<0.04 Pa s−1) is observed over the top end of Australia and the Gulf of Carpentaria (Fig. 6b), a clear indication of the active phase of the Australian monsoon trough as the deep convection embedded in the active MJO passes across the region. However, consistent with the other atmospheric parameters described in previous paragraphs, positive anomalies (descents) are located over Sulawesi, Borneo, and the central region of New Guinea, as shown in Fig. 6b. During suppressed days, major ascents are seen over the southwestern region of Sumatra, Sulawesi, and the central parts of New Guinea (Figs. 6c and 6d). During the weak regime, the anomaly pattern shows that weak upward motion occurs over Java and some parts of Borneo (Fig. 6e). Interestingly, slightly stronger ascent is seen over eastern Australia and the sea north of New Britain in the weak regime than in other regimes of the MJO.

The effect of the MJO can also be seen on thermodynamic conditions (Fig. 7). Here, the midtropospheric relative humidity is composited and averaged over 700–400 hPa. The atmospheric instability is computed from differences in pseudo equivalent potential temperature (Bolton 1980) between 850 and 500 hPa. During the active MJO, the atmosphere is humid (>60% relative humidity) over the islands of the MC but also relatively stable, except on the northeastern side of New Guinea where enhanced instability is observed (Fig. 7a). In contrast, the atmosphere is unstable over the islands during the suppressed and weak regimes (Figs. 7b and 7c). Although the thermodynamic instability is highly enhanced during the suppressed phase over Australia, associated with strong heating of the landmass, a lack of moisture reduces the rainfall over inland Australia compared to that seen for the active MJO.

Fig. 7.

The mean thermodynamic instability (difference of pseudo equivalent potential temperature between the 850- and 500-hPa levels; K), and the mean midtropospheric (average of 700–400 hPa) relative humidity (%) for the three MJO categories: (a) active, (b) suppressed, and (c) weak.

Fig. 7.

The mean thermodynamic instability (difference of pseudo equivalent potential temperature between the 850- and 500-hPa levels; K), and the mean midtropospheric (average of 700–400 hPa) relative humidity (%) for the three MJO categories: (a) active, (b) suppressed, and (c) weak.

4. Influence of the MJO on the diurnal cycle

To measure the impact of the MJO on the amplitude of the diurnal cycle of rainfall, the normalized relative amplitude (NRA) of the diurnal cycle is calculated. This is defined as the mean rainfall during 0000–1100 LT (hereafter morning rain) minus that during 1200–2300 LT (hereafter evening rain), then divided by the climatological-average rainfall for 10 DJF seasons. Figure 8 shows the horizontal distribution of the NRA for the total rainfall and for the respective three categories of the MJO. The general features of the diurnal cycle [i.e., more rainfall in the evening (morning) over land (coastal/ocean) regions] are clear in all NRA maps. The NRA has its largest size over the islands near regions of steep topography and in near-coastal ocean areas adjacent to the larger islands of the MC. Note that there is a distinct variation in the magnitude of the NRA during the three MJO regimes. For example, during active days, the NRA for regions of morning rainfall maximum (red areas; e.g., the near-coastal oceans regions to the northeast of Borneo, east of Sulawesi, northwest of Australia, and north and northeast of New Guinea) is more than 1.5 times the seasonal mean, whereas during suppressed days, it is much closer to the climatological average in many of these regions (Figs. 8b and 8c). Similarly, the magnitude of morning rainfall is greater during the active regime over the Java Sea, Timor Sea, and the Gulf of Carpentaria than in the suppressed phase of the MJO. Evening rainfall (negative NRA; blue regions) dominates during the suppressed days of the MJO over Sulawesi, the southern coast of Borneo, and mountainous regions of New Guinea. In contrast, the southern interior of Borneo and New Guinea experiences slightly more rain in the morning during suppressed days than in active days (Fig. 8c). During the weak cycle (Fig. 8d), morning rainfall maxima are noted over the sea just northeast of Australia, which corresponds to an area of larger ascent (Fig. 6d) than in other regimes in this location.

Fig. 8.

NRA for (a) the climatological-average rainfall, and (b) active, (c) suppressed, and (d) weak categories of the MJO. The NRA is calculated by subtracting the average of evening rainfall (1200–2300 LT) from morning (0000–1100 LT), which is then normalized by the climatological average of rainfall for 10 DJF seasons.

Fig. 8.

NRA for (a) the climatological-average rainfall, and (b) active, (c) suppressed, and (d) weak categories of the MJO. The NRA is calculated by subtracting the average of evening rainfall (1200–2300 LT) from morning (0000–1100 LT), which is then normalized by the climatological average of rainfall for 10 DJF seasons.

To analyze the regional variations in both the amplitude and phase of the diurnal cycle of rainfall in a concise way, an EOF analysis similar to that of Kikuchi and Wang (2008) is applied. The left (right) column of Fig. 9 shows the results from the EOF analysis of the diurnal cycle of total rainfall composited from 3B42 (3G68). Clearly, the significance test of eigenvalues (Figs. 9a and 9e), as outlined in section 2b, reveals that neighboring EOFs do not contaminate the first two modes of the EOFs. More than 86% of the total variance is explained by these two modes for 3B42 and 67% for 3G68. Thus, only the results from these first two modes are discussed. The corresponding time series of the principal components (PC1 and PC2) of the EOFs show a marked diurnal cycle (Figs. 9b and 9f). Although the spatial pattern of 3G68 is coarser (Figs. 9g and 9h), its results are consistent with 3B42 except that both of its PCs time series lead 3B42 by 2–3 h (cf. Figs. 9b and 9f). This lead also appears in the time series of PCs of the MJO phases for 3B42 and 3G68 (Figs. 11 and 12). The lead is due to the fact that one dataset (3G68) is a direct observation of rainfall, whereas the other (3B42) is estimated using IR to determine the cloud-top temperature, to increase its spatial coverage. The time of occurrence of maximum rainfall in IR-estimated rainfall is strongly affected by cloud cover and tends to lag behind in situ observations by approximately 3 h (Kubota and Nitta 2001). Modeling results also give this lag. Sato et al. (2009) found that the phase difference between rainfall and OLR as simulated in NICAM is consistent with the phase difference between the 3G68 and 3B42 products. Kikuchi and Wang (2008) have observed a similar tendency and suggested for best results using the time series of the PCs from 3G68 because of its higher temporal sampling but using the spatial pattern from 3B42 because of its higher horizontal resolution. Therefore, hereafter we have interpreted our results by using the time series of PCs from 3G68 and the spatial pattern of EOFs from 3B42. For completeness, we have also included the time series of PCs from both datasets.

Fig. 9.

EOF analysis of the diurnal cycle of climatological-average rainfall (DJF) from TRMM (left) 3B42 and (right) 3G68 data (a),(e) eigenvalue spectrum, (b),(f) time series of PC1 and PC2, and the spatial pattern of (c),(g) EOF1 and (d),(h) EOF2. The significance test based on North et al. (1982) is shown as error bars, and white lines are contours of 1000-m topographic height.

Fig. 9.

EOF analysis of the diurnal cycle of climatological-average rainfall (DJF) from TRMM (left) 3B42 and (right) 3G68 data (a),(e) eigenvalue spectrum, (b),(f) time series of PC1 and PC2, and the spatial pattern of (c),(g) EOF1 and (d),(h) EOF2. The significance test based on North et al. (1982) is shown as error bars, and white lines are contours of 1000-m topographic height.

Fig. 11.

As in Fig. 10, but for the diurnal cycle of rainfall for the suppressed category of the MJO.

Fig. 11.

As in Fig. 10, but for the diurnal cycle of rainfall for the suppressed category of the MJO.

Fig. 12.

As in Fig. 10, but for the diurnal cycle of rainfall for the weak category of the MJO.

Fig. 12.

As in Fig. 10, but for the diurnal cycle of rainfall for the weak category of the MJO.

The EOF1 for both datasets (Figs. 9c and 9g), which alone explains 62% (42%) of the total variance, shows a clear land–sea contrast in rainfall, reflecting their different thermal response to solar heating. PC1 shows the general characteristic of diurnal rainfall having a peak time in early evening over land but in the morning over the ocean. The EOF2 (Figs. 9d and 9h) explains 26% (22%) of the total variance and the corresponding spatial pattern can be interpreted as showing how the general features of diurnal rainfall are modified because of geographical location, size of an island, and the orientation of topography. Figures 9c and 9d show that the diurnal signal is stronger over the islands, tropical Australia, and surrounding near-coastal ocean areas because the magnitude of the eigenvectors in both EOFs is greater over those regions. In contrast, over open oceans and subtropical regions of Australia, the diurnal signal is weaker. For the best interpretation of the EOF results, the occurrence of daily maximum rainfall is divided into different time periods as shown in Table 2. For example, the maximum rainfall occurs between 1800 and 0000 LT when EOF1 is positive and EOF2 is negative; between 0000 and 0600 LT when both EOFs are negative; between 0600 and 1200 LT when EOF1 is negative and EOF2 is positive; and between 1200 and 1800 LT when both EOFs are positive. Based on the above definition, Fig. 9 shows near-coastal ocean regions have negative (positive) values in EOF1 (EOF2) resulting in a morning rainfall maximum (0600–1200 LT), while the larger islands of the MC have positive (negative) values in EOF1 (EOF2), resulting in a late evening rainfall maximum (1800–0000 LT). In contrast, over the small islands and the top end of Australia, both EOFs have the same sign, resulting in maximum rainfall during afternoon/early evening (1200–1800 LT). Farther poleward over the Australian continent, the peak rainfall time is delayed as late as midnight. In addition, the magnitude of EOFs and PCs also need to be considered to perceive the life cycle of rainfall systems. For example, over small islands like Java, both EOF1 and EOF2 are positive (Figs. 9c and 9d), but Fig. 9f shows that PC1 and PC2 have zero magnitude at 1300 and 1800 LT, respectively, while they are positive between these times. Combining the times series and EOF patterns (e.g., PC1 × EOF1 + PC2 × EOF2), it is found that the rainfall starts to develop in the afternoon around 1300 LT over Java, reaches its maximum at 1500–1600 LT, and starts to diminish in the early evening after 1800 LT, which is consistent with previous studies (e.g., Qian 2008).

Table 2.

Time of occurrence of the maximum rainfall according to the signs of the EOFs and PCs.

Time of occurrence of the maximum rainfall according to the signs of the EOFs and PCs.
Time of occurrence of the maximum rainfall according to the signs of the EOFs and PCs.

A 6-h (quadrature) phase difference appears when comparing the results of EOF1 and EOF2 in Fig. 9. The EOFs change sign over coastal and mountainous areas, indicating rainfall propagation both offshore and inland. For example, careful comparison of the sign of EOF1 and EOF2 over New Guinea and its adjacent sea regions (refer to Table 2) shows that rainfall starts to develop over the adjacent sea after 1300 LT, then propagates onshore. The peak in the diurnal cycle occurs during 1700–1900 LT over the mountains while peaks over the plains on both sides of the mountains occur between 2300 and 0100 LT. The rainfall then starts to propagate offshore where the maximum rainfall occurs during 0400–0600 LT (see also Fig. 15a). These results are consistent with previous studies. Kikuchi and Wang (2008) found such propagation for total rainfall over many coastal regions of the MC. According to their results, in the landward (seaward) region of the coast, the rainfall begins along the coastline from morning to noon (afternoon to midnight), propagates inland (offshore), and ceases in the evening (early afternoon). A number of previous studies have also documented similar rainfall propagation over individual islands of the MC using different datasets and for different seasons, for example, Mori et al. (2004) over Sumatra and Ichikawa and Yasunari (2006, 2008) over Borneo/New Guinea.

Fig. 15.

As in Fig. 13, but for a cross section across New Guinea.

Fig. 15.

As in Fig. 13, but for a cross section across New Guinea.

The EOF analysis is performed for all three categories of the MJO (Figs. 10 –12). In all three categories, more than 85% of the total variance is explained by the first two EOFs. Despite the geographical distribution of EOFs during all MJO phases being more or less similar to the EOFs for climatological-average rainfall, recognizable lags and leads are shown in the time series of the PCs for active, suppressed, and weak categories of the MJO compared with the PCs of total rainfall. This delay and lead appear in both datasets (3B42 and 3G68) but are more discernable in 3G68. A clear delay of 1–2 h in the time series of PC1 is observed during the active phase relative to the total rainfall (cf. Figs. 10f and 9f). Similarly, the PC1 of the suppressed (Fig. 11f) and weak regimes (Fig. 12f) of the MJO leads that of total rainfall (Fig. 9f) by 1–2 h. During suppressed/weak regimes, high convective available potential energy (CAPE) combined with local moisture convergence from sea breezes results in strong convection in coastal areas. This is responsible for early evening peaks in rainfall during those regimes. During active days, the presence of a synoptic-scale cloud likely suppresses the local convection associated with sea breezes and delays the afternoon peak in rainfall. Further investigation by regional-scale modeling is required to better understand these mechanisms.

Fig. 10.

[top (bottom)] The EOF analysis of the diurnal cycle of rainfall for the active category of the MJO from TRMM 3B42 (3G68) data: the distributions of (a),(d) EOF1, (b),(e) EOF2, and (c),(f) the time series of PCs.

Fig. 10.

[top (bottom)] The EOF analysis of the diurnal cycle of rainfall for the active category of the MJO from TRMM 3B42 (3G68) data: the distributions of (a),(d) EOF1, (b),(e) EOF2, and (c),(f) the time series of PCs.

The life cycle of rainfall over Sumatra and its offshore regions further illustrate this lead or lag in the maximum rainfall for the three MJO categories. Figure 13 shows a succinct summary of rainfall propagation over Sumatra, in the form of a Hovmöller diagram calculated from the 3B42 data over the cross-sectional area indicated in Fig. 1a. Also indicated in this figure are the corresponding 3G68 times, which as previously indicated are earlier than those for 3B42. During active days, there is strong rainfall development over the higher topography and regions immediately adjacent to the west, starting at about 1200 LT. After about 2100 LT, there is strong propagation both eastward and westward. Maximum rainfall amounts occur offshore to the west. In contrast, Fig. 13c shows that during the suppressed regime, convection starts a little earlier, after 1100 LT, just west of the mountains. Propagation to the west is still observed but propagation to the east is much weaker than during active days. Maximum rainfall amounts occur over the mountains and just to the west of them rather than over the oceans as in active days. During the weak regime (Fig. 13d), the time evolution is similar to climatology (Fig. 13a). Mori et al. (2004) also found a rainfall peak propagating from the southwestern coastline of Sumatra toward an offshore region of the Indian Ocean during 0000–1200 LT. For comparison with our results, based on the RMM index, during the intensive observation period (IOP) of Mori et al. (2004), the suppressed (weak) category of the MJO existed only for 3 (3) days and the remaining 24 days were MJO active. Therefore, their results for the migration pattern of rainfall are most comparable to our active phase results. The land area of Sumatra Island is largely rain free between 0400 and 1100 (0200 and 0900) LT during the active (suppressed/weak) regime of the MJO.

Fig. 13.

Composite Hovmöller diagram of 3B42 rainfall averaged along the rectangular domain shown in Fig. 1a (a) for total rainfall and for the three categories of the MJO: (b) active, (c) suppressed, and (d) weak. Rainfall rates greater than 0.2 mm hr−1 are shaded, and the contour interval is 0.2 mm hr−1. The dashed line at the bottom of each panel is the average elevation along the cross section, and the line with circles is the daily accumulated rainfall. The corresponding local time for TRMM 3G68 is shown in parentheses.

Fig. 13.

Composite Hovmöller diagram of 3B42 rainfall averaged along the rectangular domain shown in Fig. 1a (a) for total rainfall and for the three categories of the MJO: (b) active, (c) suppressed, and (d) weak. Rainfall rates greater than 0.2 mm hr−1 are shaded, and the contour interval is 0.2 mm hr−1. The dashed line at the bottom of each panel is the average elevation along the cross section, and the line with circles is the daily accumulated rainfall. The corresponding local time for TRMM 3G68 is shown in parentheses.

In contrast to the larger islands of the MC, over the smaller islands and land regions, such as parts of the Malay Peninsula, Java, Timor, and New Britain, both EOF1 and EOF2 tend to have positive values (Fig. 14). For example, on Java, Fig. 14 shows that during the active (suppressed) regime, maximum rainfall occurs between 1700 and 1900 (1400 and 1600) LT but between 1500 and 1700 LT during the weak cycle (not shown). Along the northern coast, rainfall maxima typically occur near or just after midnight, while just offshore, maxima occur during the morning hours. In the eastern Java Sea, between Java and Borneo, there is an anomalous region indicated by negative EOF1 and positive EOF2 values. During MJO active days, this implies a rainfall maximum between 1100 and 1300 LT, unusual for an oceanic region. During suppressed days, this feature is shifted westward. The location of this feature, situated as it is between several islands, suggests that its cause may be related to interaction between strong land and sea breezes originated in the surrounding islands. This type of interaction was simulated in this region by Qian (2008), using a regional model implemented over Java. His simulations indicate that this morning maximum over the Java Sea is likely caused by a combination of propagation of disturbances away from the islands and convergence between land breezes from Java and Borneo. During active days, this feature is located farther toward the east, which may be related to the influence of the strong synoptic-scale westerly wind that occurs here in the active regime (Fig. 6a). In contrast, during the suppressed and weak regimes, convergence may occur earlier because of less synoptic-scale eastward deflection of the converging land breezes, thus allowing convergence to occur closer to the islands. This gives an earlier diurnal rainfall peak in this region during these regimes (cf. Fig. 10f to Figs. 11f and 12f). However, gravity waves driven by diurnal heating of the elevated land surface over islands may also play a role in the diurnal propagation of rainfall over land out into to the adjacent oceans, as noted by Mapes et al. (2003) and Zuidema (2003).

Fig. 14.

Detail plot over the western MC of EOF of rainfall patterns for (a),(b) active and (c),(d) suppressed categories.

Fig. 14.

Detail plot over the western MC of EOF of rainfall patterns for (a),(b) active and (c),(d) suppressed categories.

New Guinea has the highest mountain ridge (>3000 m) among the islands of the MC. It is observed that the diurnal cycle of rainfall is pronounced not only over the island itself but also over the adjoining seas off the northern coast and in between New Guinea and Australia. However, the areal extent and phase of the diurnal distribution of rainfall over those regions are different during the three categories of MJO. During suppressed and weak days (Figs. 11a,b and 12a,b), the smaller amplitude of EOFs over the sea surrounding New Guinea indicates minimal rainfall with very weak diurnal variation occurring. In contrast, a stronger diurnal signal with greater rainfall is observed for the active days over the northern and northeastern coastal regions and the ocean area between northern Australia and the southern coast of New Guinea (Fig. 10; high negative values in EOF1 and high positive values in EOF2). Figure 15 gives the typical diurnal variation of the position of 3B42 rainfall maxima over a cross section of New Guinea given in Fig. 1a. The total rainfall pattern (Fig. 15a) shows that rainfall usually originates over the region of highest topography or just to the west of it, just after 1500 LT (here, for the reasons mentioned previously, we take the time variation from 3G68). The weather systems then typically propagate slowly away from the mountains in both directions, reaching both coastlines some time between 0300 and 0700 LT. Propagation appears stronger toward the west than the east, though, and it is not completely clear from the total rainfall pattern whether the peak near 144°E off the northeast coast after 0300 LT is caused by propagation or in situ development. Certainly, though, even if in situ development is occurring in this location, propagation of rainfall anomalies away from the coastline from this region continues into the morning hours.

Similar propagation occurs in the different regimes of the MJO (Figs. 15b–d), but its specific characteristics vary considerably between regimes. Comparing the suppressed and active regimes, during the suppressed regime, rainfall rates near the regions of high topography are more intense. Rainfall maxima during the suppressed category are farther from the topography and occur later, and the westward propagation of anomalies is much more pronounced than during the active regime. Propagation to the east appears less than for total rainfall. In the active regime, the eastern maximum near 144°E after 0300 LT is much more pronounced and there is some eastward propagation from the mountains. In contrast, in the regions off the southwest coast near 136°E, there is much more development of precipitation after 0100 LT than in the suppressed regime, resulting from a combination of what appears to be eastward propagation from the sea, in situ development, and perhaps some residual westward propagation from the land. During the weak regime (Fig. 15d), there is strong development close to the mountains, with both westward and eastward propagation, but formation over ocean regions is less than in the active regime and more similar to climatology. In addition to offshore propagation, there is some onshore propagation in the late morning in all categories, from about 136° to 140°E. This is most pronounced in the active regime and weakest in the suppressed regime.

These findings are consistent with the results of previous studies in this location (Zhou and Wang 2006; Ichikawa and Yasunari 2008). Using a high-resolution regional atmospheric model, Zhou and Wang (2006) found that orography increases the moisture convergence at low levels by blocking and deflecting the mean flow. They also noticed stronger valley (upslope) winds help to develop deep convection over mountaintops during the afternoon due to elevated topography, which acts as a source of propagating gravity waves and helps initiate rainbands in the coastal regions offshore during the early morning. However, this does not explain the inhibition of northeastward migration of rainfall near the north coast of New Guinea during the suppressed days of the MJO, as shown in Fig. 15c. To explain rainfall initiation, intensification, and migration (onshore and offshore) during easterly and westerly regimes over and around New Guinea, Ichikawa and Yasunari (2008) showed that large-scale conditions act to suppress convection over the northern coastal regions of the island during the easterly (suppressed) regime and this suppression acts as a barrier, preventing the initiation of the sea breeze that acts to intensify convection. In addition, it is likely that land breezes are not so strong in this regime and thus they do not aid the propagation of rainfall northeastward. We also notice a similar feature (Figs. 5b and 6b) over the northern coastal regions of New Guinea during suppressed days of the MJO. Recent results (e.g., Zhou and Wang 2006; Qian 2008; Wu et al. 2009) have shown that high-resolution cloud-resolving models are able to produce realistic characteristics of the diurnal cycle of rainfall over the islands of the MC. Hence, evaluating a model-simulated diurnal cycle during different phases of the MJO would shed more light on the mechanisms of such characteristics of the diurnal variation.

Large differences in rainfall building processes are also noted over the tropical regions of Australia. Over the inland regions of northern Australia between 12° and 14°S and 130° and 135°E, large-scale convection starts to develop after 1300 LT (cf. Fig. 10b, positive values of EOF2 and Fig. 10f, positive maximum in PC2) and prominent rainfall maxima occur between 1900 and 2100 LT during active days (Fig. 10a, positive values of EOF1 and Fig. 10f, positive maximum in PC1). This then expands farther south and southwest, so that south of 16°S, maximum rainfall occurs after midnight (Fig. 10b, negative values of EOF2 and Fig. 10f, negative maximum in PC2). After 0100–0300 LT, rainfall starts to develop just off the western coastlines (Figs. 10b and 10f) and reaches its maximum between 0600 and 0900 LT over the Timor Sea (Figs. 10a and 10f). In contrast, during suppressed and weak days, early initiation of rainfall between 0000 and 0200 LT (Figs. 11f and 12f; PC2) is observed along the western coastline near the top end (Figs. 11b and 12b), mostly concentrated over the coastal regions (Figs. 11a and 12a) where the maximum rainfall occurs between 0200 and 0400 LT (Figs. 11f and 12f; PC1). Isolated rainfall systems develop over the Tiwi Islands and the top end of Australia (Fig. 11a) in the early afternoon after 1200 LT (Fig. 11f; PC2) during the suppressed days of MJO, which then intensify, merge and, based on similar observations from Pope et al. (2008), likely are related to organized mesoscale convective systems (MCSs) that often occur over the top end of Australia near Darwin, with maximum rainfall between 1600 and 1700 LT (Figs. 11a and 11f). Figure 11a shows that during the suppressed regime, southward propagation is limited to 16°S, where rainfall ceases in the early evening. Analysis of the in situ rainfall at Darwin Airport, stratified by the RMM index, shows that the suppressed regime contributes more rainfall during the afternoon (1700 LT), while during the active regime of the MJO, more rainfall occurs during the early morning (0300 LT), but a secondary maximum is also noted in the afternoon during active days (Rauniyar and Walsh 2009). This is consistent with the results of Pope et al. (2008), who observed that the diurnal cycle of the number of mean cluster elements (CE; a patch of cloudy pixels) over land peaks in the afternoon but reaches its maximum extent several hours later. On stratifying the results according to the zonal component of the 700-hPa wind, they observed that the diurnal cycle of mean CE area has unimodal (bimodal) distribution for easterly (westerly) wind regimes with a peak near 2200 (1600 and 0400) LT. Their results reveal a nocturnal (afternoon) MCSs genesis peak and also that MCSs genesis and decay resemble those of other oceanic (continental) areas during active (suppressed) monsoon conditions over Darwin.

Over Cape York Peninsula, a large area is covered with rainfall during active (suppressed/weak) days and the rainfall rate peaks between 1900 and 2100 (1600 and 1800) LT. The diurnal cycle of rainfall is weak over the subtropical (inland) regions of Australia during all phases of the MJO. In contrast, Fig. 10d (Fig. 10e) shows high positive (negative) values in EOF1(2) over the region 18°–20°S, 129°–131°E during the MJO active phase. Combining the corresponding time series (Fig. 10f) of PC1 and PC2, it is found that short-lived rainfall (6 h) occurs in this location that starts at 1700–1800 LT and ceases after 0000 LT, which is not seen during other MJO phases. Strong moisture convergence (Fig. 5a) combined with an unstable atmosphere (Fig. 7a) is seen in that area during active days, suggesting that these might be contributing factors. However, results from a high-resolution regional-scale model will be required to further understand the causes of such a peculiar diurnal cycle in this inland area of Australia. Significant differences in rainfall-building processes around the Gulf of Carpentaria are seen between MJO phases, where during active/weak days (Figs. 10 and 12) the maximum rainfall occurs between 0900 and 1200 LT, almost 3–6 h after the peak during suppressed days (Fig. 11). During suppressed days, upward motions are suppressed (Fig. 6b) and also the migration of convection from regions south of New Guinea may be resisted by a synoptic-scale westerly wind component.

5. Summary and discussion

This paper presents a comprehensive study on the characteristics of the diurnal cycle of rainfall during different phases of the MJO over the MC by utilizing a high-resolution TRMM dataset. This study has found distinct regional variations in the rainfall pattern among phases of the MJO over land and ocean. The composite-mean rainfall patterns for DJF show that the average daily rainfall rate over islands of the MC is higher during suppressed MJO days. This behavior is opposite for northern regions of Australia where more rainfall occurs during MJO active days. Distinct variations in large-scale atmospheric dynamics are also observed during the phases of MJO. The atmosphere is more humid during the active MJO regime but found to be more stable thermodynamically, with much less upward motion resulting in less rainfall over the islands of the MC. In contrast, more rainfall than average occurs over northern Australia during the active regime, a consequence of strong moisture convergence associated with the Australian monsoon. However, the rainfall over the sea is most affected by the phases of MJO. Over the ocean regions, more rainfall occurs during the active regime off the southern coast of Sumatra, the Java, Timor, and Arafura Seas and over the ocean region north and south of New Guinea. These rainfall patterns were found to be largely consistent with OLR anomalies and with regions of strong moisture convergence. During the suppressed regime, positive rainfall anomalies are smaller, associated with strong vertical instability, and are concentrated over the larger MC islands, consistent with the results of Ichikawa and Yasunari (2006, 2008), who also found that in their easterly regime (i.e., suppressed days) convective activity is centered over the western MC around Sumatra, Borneo, and Sulawesi.

A number of studies have documented the diurnal cycle of total rainfall over individual islands of the MC, but very few have examined the effect of intraseasonal variations on the amplitude and phase of the diurnal cycle of rainfall. Our results for the diurnal cycle of total rainfall are consistent with earlier findings and also strengthen the conclusions of Kikuchi and Wang (2008). However, Kikuchi and Wang (2008) did not examine the intraseasonal modulation of the diurnal cycle. Our analysis of the normalized relative amplitude (NRA) of the diurnal cycle shows that its amplitude is more than one and a half times the climatological-average diurnal cycle amplitude over near-coastal ocean regions during active days of the MJO while the amplitude is less than the climatological average during other phases of the MJO. Similarly, the NRA is greater than average in the regions of the islands of the MC that have an evening rainfall maximum during suppressed days of the MJO.

The EOF analysis presented here is able to determine the characteristics of the diurnal cycle, with its first two modes alone explaining more than 86% of the total variance. A clear land–sea contrast across the entire domain appears in the first mode of the EOFs, while the second EOF mode represents the features whose diurnal cycle is affected by the interaction of land–sea boundaries and elevated topography. The corresponding PC time series show a marked diurnal cycle. The EOF analysis of the diurnal cycle of rainfall for each MJO regime shows how the MJO modulates the amplitude and phase of the diurnal cycle of rainfall. In general, the peak time of diurnal rainfall during the active (suppressed) regime of MJO is delayed (led) by 2 h compared to the diurnal cycle of total rainfall. In addition, the rainfall propagation pattern is also modulated by the regime of the MJO. During the active regime, the rainfall propagates several hundred kilometers away from the coastal regions toward open sea, while it is limited to near coastal areas in other regimes. At Darwin, it has been found that the active phase of the MJO is responsible for enhancement of nighttime rainfall over inland regions and nearby small islands but a secondary maximum is also noted in the afternoon during active days (Rauniyar and Walsh 2009). The probable explanation for this is the presence of large-scale cloudiness retarding the initiation of sea breezes during active days. Farther inland over Australia, in the region 18°–20°S, 129°–131°E, the active phase of the MJO is responsible for short-lived evening rainfall. This unusual feature is accompanied by strong moisture convergence at 850 hPa along with a thermodynamically unstable midatmosphere, but the precise mechanisms for the enhanced rainfall in this location remain unclear. Further investigation of these mechanisms may be possible through numerical modeling.

This paper has investigated the impact of the MJO on the diurnal cycle of rainfall, but it is possible that, because this is a scale-interaction phenomenon, there may also be some influence of the diurnal cycle on the MJO. For example, Raupp and Silva Dias (2009) show that the diurnal variation in convection can modulate intraseasonal oscillations, through the generation of inertia–gravity waves by diurnal convection. The present study did not address this issue, although fine-resolution modeling could be applied to this problem to understand better the scale interaction of diurnal convection and the MJO in the MC. Similarly, it has been suggested that the equatorial two-day wave (Chen and Houze 1997), which is most prevalent during the active regime of the MJO, is a westward-propagating inertia–gravity wave (Haertel and Kiladis 2004). Thus, it may also have a relationship to the modulation of the MJO. The present analysis could be extended to examine the relationship between the different categories of the MJO and the amplitude of two-day rather than diurnal oscillations.

The diurnal cycle of convection over the islands of the MC during the suppressed phase of the MJO may also be related to the establishment of the large-scale conditions required to generate the MJO. Earlier results from the Tropical Ocean Global Atmosphere (TOGA) Coupled Ocean–Atmosphere Response Experiment (COARE) showed the importance of cumulus congestus clouds, with tops that terminate around the freezing level at 0°C level, and pointed to a trimodal cloud distribution consisting of congestus, shallow, and deep convection (Johnson et al. 1999). Cumulus congestus are most prevalent during the suppressed phase of the MJO, indicating that they may serve to moisten the midtroposphere and create favorable conditions for the upcoming active regime of the MJO. Subsequent to the active regime, the development of a stable layer around the freezing level, which is frequently observed over the tropical oceans, may act to reinforce the transition from an enhanced convective phase to the suppressed regime of the MJO (Slingo et al. 2003). The present study does not address this issue, as the analysis here has focused on deep convection as a cause of total rainfall amounts, but it could be extended to detect the presence of cumulus congestus cloud through its OLR signature and examine its contribution to the relationship between the diurnal cycle and the phase of the MJO.

Acknowledgments

The authors wish to thank Prof. Ian Simmonds, Dr. Matthew Wheeler, and Dr. Peter May for their valuable suggestions and comments. We are very grateful for constructive comments from three anonymous reviewers, who greatly improved the content of this manuscript.

REFERENCES

REFERENCES
Bolton
,
D.
,
1980
:
The computation of equivalent potential temperature.
Mon. Wea. Rev.
,
108
,
1046
1053
.
Chang
,
C-P.
,
P. A.
Harr
, and
H-J.
Chen
,
2005
:
Synoptic disturbances over the equatorial South China Sea and western Maritime Continent during boreal winter.
Mon. Wea. Rev.
,
133
,
489
503
.
Chen
,
S. S.
, and
R. A.
Houze
Jr.
,
1997
:
Diurnal variation and life-cycle of deep convective systems over the tropical Pacific warm pool.
Quart. J. Roy. Meteor. Soc.
,
123
,
357
388
.
Dai
,
A.
,
2001
:
Global precipitation and thunderstorm frequencies. Part II: Diurnal variations.
J. Climate
,
14
,
1112
1128
.
Gray
,
W. M.
, and
R. W.
Jacobson
,
1977
:
Diurnal variation of deep cumulus convection.
Mon. Wea. Rev.
,
105
,
1171
1181
.
Haertel
,
P. T.
, and
G. N.
Kiladis
,
2004
:
Dynamics of 2-day equatorial waves.
J. Atmos. Sci.
,
61
,
2707
2721
.
Hannachi
,
A.
,
I. T.
Jolliffe
, and
D. B.
Stephenson
,
2007
:
Empirical orthogonal functions and related techniques in atmospheric science: A review.
Int. J. Climatol.
,
27
,
1119
1152
.
Hara
,
M.
,
T.
Yoshikane
,
H. G.
Takahashi
,
F.
Kimura
,
A.
Noda
, and
T.
Tokioka
,
2009
:
Assessment of the diurnal cycle of precipitation over the Maritime Continent simulated by a 20 km mesh GCM using TRMM PR data.
J. Meteor. Soc. Japan
,
87A
,
413
424
.
Hartmann
,
D. L.
, and
H. H.
Hendon
,
2007
:
Atmospheric science: Resolving an atmospheric enigma.
Science
,
318
,
1731
1732
.
Hidayat
,
R.
, and
S.
Kizu
,
2010
:
Influence of the Madden–Julian Oscillation on Indonesian rainfall variability in austral summer.
Int. J. Climatol.
,
30
,
1816
1825
.
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
,
38
55
.
Ichikawa
,
H.
, and
T.
Yasunari
,
2006
:
Time and space characteristics of diurnal rainfall over Borneo and surrounding oceans as observed by TRMM-PR.
J. Climate
,
19
,
1238
1260
.
Ichikawa
,
H.
, and
T.
Yasunari
,
2008
:
Intraseasonal variability in diurnal rainfall over New Guinea and the surrounding oceans during austral summer.
J. Climate
,
21
,
2852
2868
.
Johnson
,
R. H.
,
T. M.
Rickenbach
,
S. A.
Rutledge
,
P. E.
Ciesielski
, and
W. H.
Schubert
,
1999
:
Trimodal characteristics of tropical convection.
J. Climate
,
12
,
2397
2418
.
Kanamitsu
,
M.
,
W.
Ebisuzaki
,
J.
Woollen
,
S-K.
Yang
,
J. J.
Hnilo
,
M.
Fiorino
, and
G. L.
Potter
,
2002
:
NCEP–DOE AMIP-II Reanalysis (R-2).
Bull. Amer. Meteor. Soc.
,
83
,
1631
1643
.
Kikuchi
,
K.
, and
B.
Wang
,
2008
:
Diurnal precipitation regimes in the global tropics.
J. Climate
,
21
,
2680
2696
.
Kraus
,
E. B.
,
1963
:
The diurnal precipitation change over the sea.
J. Atmos. Sci.
,
20
,
551
556
.
Kubota
,
H.
, and
T.
Nitta
,
2001
:
Diurnal variations of tropical convection observed during the TOGA-COARE.
J. Meteor. Soc. Japan
,
79
,
815
830
.
Kummerow
,
C.
,
W.
Barnes
,
T.
Kozu
,
J.
Shiue
, and
J.
Simpson
,
1998
:
The Tropical Rainfall Measuring Mission (TRMM) sensor package.
J. Atmos. Oceanic Technol.
,
15
,
809
816
.
Lau
,
W. K. M.
, and
D. E.
Waliser
,
2005
:
Intraseasonal Variability in the Atmosphere–Ocean Climate System.
Praxis, 436 pp
.
Liebmann
,
B.
, and
C. A.
Smith
,
1996
:
Description of a complete (interpolated) outgoing longwave radiation dataset.
Bull. Amer. Meteor. Soc.
,
77
,
1275
1277
.
Lin
,
J-L.
, and
Coauthors
,
2006
:
Tropical intraseasonal variability in 14 IPCC AR4 climate models. Part I: Convective signals.
J. Climate
,
19
,
2665
2690
.
Lorenz
,
E. N.
,
1956
:
Empirical Orthogonal Functions and Statistical Weather Prediction.
MIT Scientific Rep. 1, Statistical Forecasting Project, Defense Document Center No. 110268, 49 pp
.
Madden
,
R. A.
, and
P. R.
Julian
,
1971
:
Detection of a 40-50 day oscillation in the zonal wind in the tropical Pacific.
J. Atmos. Sci.
,
28
,
702
708
.
Madden
,
R. A.
, and
P. R.
Julian
,
1994
:
Observations of the 40-50-day tropical oscillation: A review.
Mon. Wea. Rev.
,
122
,
814
837
.
Mapes
,
B. E.
,
T. T.
Warner
, and
M.
Xu
,
2003
:
Diurnal patterns of rainfall in northwestern South America. Part III: Diurnal gravity waves and nocturnal convection offshore.
Mon. Wea. Rev.
,
131
,
830
844
.
Massie
,
S. T.
,
J.
Gille
,
C.
Craig
,
R.
Khosravi
,
J.
Barnett
,
W.
Read
, and
D.
Winker
,
2010
:
HIRDLS and CALIPSO observations of tropical cirrus.
J. Geophys. Res.
,
115
,
D00H11
.
doi:10.1029/2009JD012100
.
Miura
,
H.
,
M.
Satoh
,
T.
Nasuno
,
A. T.
Noda
, and
K.
Oouchi
,
2007
:
A Madden–Julian Oscillation event realistically simulated by a global cloud-resolving model.
Science
,
318
,
1763
.
Mori
,
S.
, and
Coauthors
,
2004
:
Diurnal land–sea rainfall peak migration over Sumatera Island, Indonesian Maritime Continent observed by TRMM satellite and intensive rawinsonde soundings.
Mon. Wea. Rev.
,
132
,
2021
2039
.
Neale
,
R. B.
, and
J. M.
Slingo
,
2003
:
The Maritime Continent and its role in the global climate: A GCM study.
J. Climate
,
16
,
834
848
.
Negri
,
A. J.
,
T. L.
Bell
, and
L.
Xu
,
2002
:
Sampling of the diurnal cycle of precipitation using TRMM.
J. Atmos. Oceanic Technol.
,
19
,
1333
1344
.
North
,
G. R.
,
T. L.
Bell
,
R. F.
Cahalan
, and
F. J.
Moeng
,
1982
:
Sampling errors in the estimation of empirical orthogonal functions.
Mon. Wea. Rev.
,
110
,
699
706
.
Ohsawa
,
T.
,
H.
Ueda
, and
T.
Hayashi
,
2001
:
Diurnal variations of convective activity and rainfall in tropical Asia.
J. Meteor. Soc. Japan
,
79
,
333
352
.
Pope
,
M.
,
C.
Jakob
, and
M. J.
Reeder
,
2008
:
Convective systems of the north Australian monsoon.
J. Climate
,
21
,
5091
5112
.
Qian
,
J. H.
,
2008
:
Why precipitation is mostly concentrated over islands in the Maritime Continent.
J. Atmos. Sci.
,
65
,
1428
1441
.
Rashid
,
H.
,
H.
Hendon
,
M.
Wheeler
, and
O.
Alves
,
2010
:
Prediction of the Madden–Julian Oscillation with the POAMA dynamical seasonal prediction system.
Climate Dyn.
,
in press, doi:10.1007/s00382-010-0754-x
.
Rauniyar
,
S.
, and
K.
Walsh
,
2009
:
Diagnosing the effect of MJO on the diurnal cycle of rainfall in the Southern Hemisphere.
Proc. Ninth Int. Conf. on Southern Hemisphere Meteorology and Oceanography, Melbourne, Australia, Australian Meteorological and Oceanographic Society, 151–154
.
Raupp
,
C. F. M.
, and
P. L.
Silva Dias
,
2009
:
Resonant wave interactions in the presence of a diurnally varying heat source.
J. Atmos. Sci.
,
66
,
3165
3183
.
Sakurai
,
N.
, and
Coauthors
,
2005
:
Diurnal cycle of cloud system migration over Sumatera Island.
J. Meteor. Soc. Japan
,
83
,
835
850
.
Sato
,
T.
,
H.
Miura
,
M.
Satoh
,
Y. N.
Takayabu
, and
Y.
Wang
,
2009
:
Diurnal cycle of precipitation in the tropics simulated in a global cloud-resolving model.
J. Climate
,
22
,
4809
4826
.
Satoh
,
M.
,
T.
Matsuno
,
H.
Tomita
,
H.
Miura
,
T.
Nasuno
, and
S.
Iga
,
2008
:
Nonhydrostatic icosahedral atmospheric model (NICAM) for global cloud resolving simulations.
J. Comput. Phys.
,
227
,
3486
3514
.
Shin
,
D. W.
,
S.
Cocke
, and
T. E.
LaRow
,
2007
:
Diurnal cycle of precipitation in a climate model.
J. Geophys. Res.
,
112
,
D13109
.
doi:10.1029/2006JD008333
.
Slingo
,
A.
,
K. I.
Hodges
, and
G. J.
Robinson
,
2004
:
Simulation of the diurnal cycle in a climate model and its evaluation using data from Meteosat 7.
Quart. J. Roy. Meteor. Soc.
,
130
,
1449
1467
.
Slingo
,
J.
, and
Coauthors
,
1996
:
Intraseasonal oscillations in 15 atmospheric GCMs: Results from an AMIP diagnostic subproject.
Climate Dyn.
,
12
,
325
357
.
Slingo
,
J.
,
P.
Inness
,
R.
Neale
,
S.
Woolnough
, and
G. Y.
Yang
,
2003
:
Scale interactions on diurnal to seasonal timescales and their relevance to model systematic errors.
Ann. Geophys.
,
46
,
139
155
.
Tian
,
B.
,
B. J.
Soden
, and
X.
Wu
,
2004
:
Diurnal cycle of convection, clouds, and water vapor in the tropical upper troposphere: Satellites versus a general circulation model.
J. Geophys. Res.
,
109
,
D10101
.
doi:10.1029/2003JD004117
.
Tian
,
B.
,
D. E.
Waliser
, and
E. J.
Fetzer
,
2006
:
Modulation of the diurnal cycle of tropical deep convective clouds by the MJO.
Geophys. Res. Lett.
,
33
,
L20704
.
doi:10.1029/2006GL027752
.
Tripoli
,
G. J.
, and
W. R.
Cotton
,
1989a
:
Numerical study of an observed orogenic mesoscale convective system. Part I: Simulated genesis and comparison with observations.
Mon. Wea. Rev.
,
117
,
273
304
.
Tripoli
,
G. J.
, and
W. R.
Cotton
,
1989b
:
Numerical study of an observed orogenic mesoscale convective system. Part II: Analysis of governing dynamics.
Mon. Wea. Rev.
,
117
,
305
328
.
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
,
1917
1932
.
Wheeler
,
M. C.
,
H. H.
Hendon
,
S.
Cleland
,
H.
Meinke
, and
A.
Donald
,
2009
:
Impacts of the Madden–Julian Oscillation on Australian rainfall and circulation.
J. Climate
,
22
,
1482
1498
.
Wu
,
P.
,
M. D.
Yamanaka
, and
J.
Matsumoto
,
2008
:
The formation of nocturnal rainfall offshore from convection over western Kalimantan (Borneo) Island.
J. Meteor. Soc. Japan
,
86A
,
187
203
.
Wu
,
P.
,
M.
Hara
,
J.
Hamada
,
M. D.
Yamanaka
, and
F.
Kimura
,
2009
:
Why a large amount of rain falls over the sea in the vicinity of western Sumatra Island during nighttime.
J. Appl. Meteor. Climatol.
,
48
,
1345
1361
.
Yang
,
G-Y.
, and
J.
Slingo
,
2001
:
The diurnal cycles in the tropics.
Mon. Wea. Rev.
,
129
,
784
801
.
Yang
,
S.
, and
E. A.
Smith
,
2006
:
Mechanisms for diurnal variability of global tropical rainfall observed from TRMM.
J. Climate
,
19
,
5190
5226
.
Yang
,
S.
,
L.
Hu
,
Y.
Li
, and
S.
Gao
,
2009
:
Precipitation diurnal variability over East Asia and its connection to ENSO.
Geophysical Research Abstracts, Vol. 11, Abstract 1392
.
Zhang
,
C.
,
2005
:
Madden–Julian Oscillation.
Rev. Geophys.
,
43
,
RG2003
.
doi:10.1029/2004RG000158
.
Zhou
,
L.
, and
Y.
Wang
,
2006
:
Tropical Rainfall Measuring Mission observation and regional model study of precipitation diurnal cycle in the New Guinean region.
J. Geophys. Res.
,
111
,
D17104
.
doi:10.1029/2006JD007243
.
Zuidema
,
P.
,
2003
:
Convective clouds over the Bay of Bengal.
Mon. Wea. Rev.
,
131
,
780
798
.

Footnotes

Corresponding author address: Surendra Rauniyar, School of Earth Sciences, The University of Melbourne, VIC 3010, Australia. Email: s.rauniyar@pgrad.unimelb.edu.au