Abstract

Observational studies suggest that the vertical structure of diabatic heating is important to MJO development. In particular, the lack of a top-heavy heating profile was believed to be responsible for poor MJO simulations in global climate models. In this work, the role of the vertical heating profile in MJO simulation is investigated by modifying the convective heating profile to different shapes, from top-heavy heating to bottom cooling, to mimic mesoscale heating in the NCAR Community Atmosphere Model, version 5.3 (CAM5.3). Results suggest that incorporating a mesoscale stratiform heating structure can significantly improve the MJO simulation. By artificially adding stratiform-like heating and cooling in the experiments, many observed features of MJO are reproduced, including clear eastward propagation, a westward-tilted vertical structure of MJO-scale anomalies of dynamic and thermodynamic fields, and strong 20–80-day spectral power. Further analysis shows an abundance of shallow convection ahead of MJO deep convection, confirming the role of shallow convection in preconditioning the atmosphere by moistening the lower troposphere ahead of deep convection during the MJO life cycle. Additional experiments show that lower-level cooling contributes more to improving the MJO simulation. All these features are lacking in the control simulation, suggesting that the mesoscale stratiform heating, especially its lower-level cooling component, is important to MJO simulation.

1. Introduction

The Madden–Julian oscillation (MJO; Madden and Julian 1971, 1972) is a fundamental tropical process contributing to intraseasonal variability. It has been widely studied because of its significant influence on tropical and global climate and weather systems (Zhang 2005). However, it is very difficult to predict MJO accurately by global weather prediction models and to simulate it realistically by global climate models (GCMs). Several possible reasons have been suggested for poor MJO simulation in models, including no expected moisture preconditioning (e.g., Zhang and Mu 2005; Hsu and Li 2012; Hsu et al. 2014), lack of shallow convection ahead of deep convection (e.g., Zhang and Song 2009), and incorrect vertical structure of heating profile (e.g., Lin et al. 2004; Fu and Wang 2009; Lappen and Schumacher 2012). Failing to understand MJO thoroughly exposes gaps in our knowledge of how the atmosphere operates in the tropics. Many studies suggested that tropical large-scale circulation is sensitive to vertical structure of diabatic heating (Hartmann et al. 1984; Schumacher et al. 2004; Li et al. 2009), and the heating distribution should be one of the critical processes in governing the atmospheric variability.

Diabatic heating is the driving force for large-scale atmospheric circulation. Therefore, it is critical to the structure and propagation of MJO ultimately. In the tropics, latent heat release from active convection, especially deep convection, is a dominant component of the total diabatic heating (Jiang et al. 2011). Previous studies have suggested that systematic errors in heating profile in the model may be responsible for the poor simulation of MJO, especially the lack of the stratiform-like heating in models (e.g., Lin et al. 2004; Fu and Wang 2009). Recently, Wang et al. (2017) analyzed the MJOs simulated by more than two dozen global climate models that participated in the MJO intercomparison project (Jiang et al. 2015) and concluded that stratiform heating in the rear of MJO convection played an important role in simulating the MJO eastward propagation in GCMs.

Using the TRMM observations, Lin et al. (2004) and Barnes et al. (2015) showed that the vertical heating profile was very top-heavy during MJO development, corresponding to heating observed in the stratiform precipitation area, with upper-tropospheric heating and lower-tropospheric cooling. The MJOs may also have a larger fraction of stratiform precipitation than the seasonal mean value (Houze 1989). At the same time, the seasonal mean heating of reanalysis products showed noticeable bottom cooling along the tropical belt (Jiang et al. 2009). Favorable conditions for deep convection followed by stratiform rain (Lin et al. 2004) may be provided by shallow convection (Mapes 2000). In addition, the stratiform fraction of total rainfall was illustrated to vary with the MJO evolution, suggesting a fundamental role of stratiform heating in the process of intraseasonal variability (Lin et al. 2004; Jiang et al. 2009). Current deep convective parameterization schemes do not include organized mesoscale convection in most models. Anber et al. (2016) suggested that vertical wind shear organizes convection and, through cloud–radiation interaction, the diabatic heating profile resembles that of stratiform structure with low-level cooling. Wing and Emanuel (2014) argue that tropical convection undergoes spontaneous organization through radiative–convective feedbacks. Other studies also showed that the bottom-heavy heating related to shallow convection is essential to the boundary moisture convergence (Wu 2003) and MJO simulation (Li et al. 2009). Observations show that shallow convective heating occurs ahead of active deep convection during MJO (Jiang et al. 2011), and when the bottom-heavy heating was included, the eastward propagation speed of MJO was simulated more accurately (Lau and Peng 1987). The importance of lower-level diabatic heating was first proposed by Li (1983). It was suggested that heating profiles peaking in the lower troposphere could produce slow wave–conditional instability of the second kind (CISK) modes (Lau and Peng 1987). In an idealized GCM experiment, artificially lowering the maximum heating level improved MJO simulation significantly (Li et al. 2009). A convective parameterization scheme that produces reasonable MJOs shows more shallow convection and the atmospheric preconditioning ahead of deep MJO convection (Slingo et al. 2003; Mu and Zhang 2006, 2008). This was attributed to the changes of heating distribution. Wu (2003) showed the relationship between shallow convection and diabatic heating from a theoretical perspective. Convection may play a role in MJO through its interactions with boundary layer moisture convergence (Lau and Peng 1987). These studies show that lower-level heating associated with shallow convection and moisture convergence acts as a precursor. MJO theories have yet to consider the variation in the heating profile during the life cycle of the intraseasonal variability (Zhang 2005).

This study investigates the role of heating profile in MJO simulation by modifying the convective heating to different shapes, from top-heavy heating to bottom cooling, to mimic mesoscale stratiform heating structure in the National Center for Atmospheric Research (NCAR) Community Atmosphere Model, version 5.3 (CAM5.3). It is found that mesoscale stratiform heating is crucial to the improvement of MJO simulation, especially the lower-level cooling component. Section 2 will describe the experiment design and briefly introduce the model and data. In section 3, we will show the results of the simulation. We will also evaluate the performance of MJO simulation and analyze the potential reasons. Additional sensitivity tests of heating and cooling profiles are performed in section 4. Section 5 will discuss and summarize the paper.

2. Method and data

a. Model and experiment design

The model used in this study is the standard version of the NCAR Community Atmosphere Model, version 5.3. Its horizontal resolution is 1.9° × 2.5° and vertical resolution is 30 levels (Neale et al. 2010). Deep convection is parameterized by the Zhang–McFarlane (ZM) scheme (Zhang and McFarlane 1995) with modifications by Neale et al. (2008) to include entrainment dilution in calculating convective available potential energy (CAPE). The ZM scheme is based on a plume ensemble approach and assumes that an ensemble of convective updrafts exists when the atmosphere in the lower troposphere is conditionally unstable. Shallow convection parameterization is from Park and Bretherton (2009). CAM5.3 is the atmospheric component of the Community Earth System Model, version 1 (CESM1).

Previous studies (e.g., Lin et al. 2004) suggest that condensational heating associated with MJO consists of convective and stratiform heating. The convective heating peaks in the midtroposphere, while stratiform heating heats the atmosphere in the upper troposphere from anvil condensation and cools the atmosphere in the lower troposphere owing to evaporative cooling of falling rain. However, none of the existing convective parameterization schemes, except one by Donner (1993), accounts for mesoscale stratiform heating. Moncrieff and Liu (2006) proposed a prototype parameterization to represent mesoscale stratiform heating through what they called a “predictor corrector” approach. First, a convective scheme is used to predict the diabatic heating profile. Then the heating profile is readjusted to incorporate a first baroclinic mode, with heating in the upper half and cooling in the lower half of the convective layer. To understand the role of diabatic heating profile in the intraseasonal oscillation associated with MJO, we follow a similar strategy and modify the vertical structure of heating profile in the ZM scheme in CAM5.3. To emulate the observed heating structure, we artificially modify the convective heating profile to make it represent the total condensational heating. We use a half-sine function to approximate convective heating and two half-sine functions to represent stratiform heating from condensation in the upper levels and stratiform cooling from rain evaporation in the lower levels.

At each time step after convective parameterization is called, we reset the vertical distribution of convective heating to the following shape:

 
formula

where A0, A1, and A2 are constants and will be specified shortly. The terms PT and PB are cloud-top and cloud-base pressure, respectively. Variable Pm is the pressure level near the midtroposphere, also specified shortly. Thus, a heating profile of any shape from the ZM scheme is fitted into three components: a half-sine wave spanning the troposphere representing deep convective heating, a half-sine wave in the upper troposphere representing stratiform condensational heating, and a negative half-sine wave in the lower troposphere representing evaporative cooling in mesoscale downdrafts. Figure 1a shows a schematic of the heating profile modification, with the red line representing convective heating, yellow line representing stratiform heating, and purple line representing evaporative cooling, joined at level Pm.

Fig. 1.

(a) Schematic showing the vertical heating profile from convective parameterization before (red) and after (green) modification. A half-sine wave shape over the deep convective layer is assumed for convective heating. Yellow and purple half-sine waves represent stratiform heating and cooling, respectively. The black dashed line with the letter γ marks the level separating stratiform heating from stratiform cooling. (b) Actual deep convective heating profiles after modification of CTRL, TOPH, BOTC, TOPH2, and BOTC2, averaged for two years (1996–97) in a region (5°S–5°N, 145°–155°E) of the western Pacific.

Fig. 1.

(a) Schematic showing the vertical heating profile from convective parameterization before (red) and after (green) modification. A half-sine wave shape over the deep convective layer is assumed for convective heating. Yellow and purple half-sine waves represent stratiform heating and cooling, respectively. The black dashed line with the letter γ marks the level separating stratiform heating from stratiform cooling. (b) Actual deep convective heating profiles after modification of CTRL, TOPH, BOTC, TOPH2, and BOTC2, averaged for two years (1996–97) in a region (5°S–5°N, 145°–155°E) of the western Pacific.

We will consider stratiform heating and cooling only when the cloud tops are above the 700-hPa level, as shallow convection does not produce stratiform anvils. Therefore, for cloud tops below 700 hPa, in Eq. (1). To ensure energy conservation, column-integrated heating before and after heating profile modification must be the same; that is, , where Q0 is convective heating produced by the ZM convection scheme, g is gravitational acceleration, and p is pressure. Thus, we have

 
formula

where .

For cloud tops above the 700-hPa height, stratiform heating will be taken into consideration. In Eq. (1), Pm is determined by , where γ represents the distance from the cloud top normalized by cloud thickness (see Fig. 1a). A smaller γ value implies a larger amplitude but a thinner layer of stratiform heating. Conversely, a larger γ value implies a stronger and shallower lower-level evaporative cooling. Integrating Eq. (1) over pressure depth from PT to PB and equating it to to satisfy energy conservation before and after heating profile modification, we have

 
formula

We assume that stratiform heating in the upper levels cancels out the evaporative cooling in the lower levels. Thus, A0 is the same as in Eq. (2), and A1 and A2 are given by

 
formula

From Lin et al. (2004), the peak level of stratiform heating is around 400 hPa. Since we consider stratiform heating only when the cloud top is higher than 700 hPa, we relate A1 to A0 by

 
formula

so that deeper convection has larger stratiform heating amplitude.

Equations (1)(5) define the diabatic heating structure to be used in this study. To understand the role of heating profile in MJO simulation, five experiments are carried out. The first one is the standard NCAR CAM5.3 run [control (CTRL)]. For the second simulation, we want to emulate a top-heavy shape of heating profile by setting γ = ⅓ (referred to as TOPH). This way, according to Eq. (4) the heating amplitude in the upper layer of convection is more than twice the cooling amplitude in the lower layer. The third simulation uses an increased lower-level cooling profile by setting γ = ½ (referred to as BOTC). After carrying out a test simulation, we noted that most of the time tropical deep convection in CAM5.3 does not penetrate deep enough into the upper troposphere. Therefore, the top-heavy heating profile of the second simulation is further modified by resetting the highest convective plume to the 168-hPa level (model level 12 in the 30-level CAM5.3) at each time step. This will force the convective heating to be deep. This fourth experiment is referred to as TOPH2. The fifth experiment increases the amplitude of lower-level cooling further by setting γ = 0.7 (referred to as BOTC2). The only difference between BOTC and BOTC2 is the value of γ. Here, unlike in TOPH2 the convective top is not artificially lifted, thus emphasizing the lower-level cooling in this experiment.

Modifications for all experiments are applied at each time step within 20°S and 20°N. Table 1 lists the parameter settings for each experiment. In this study, upper-level heating refers to heating above 500 hPa, lower-level cooling refers to cooling below 700 hPa, stratiform-like heating refers to the structure of upper-level heating from anvil condensation and lower-level cooling from rain evaporation at the same time, and top-heavy structure refers to the shape of heating profile peaking in the upper troposphere. The actual heating profiles from CAM5.3 after modification of CTRL, TOPH, BOTC, TOPH2, and BOTC2 are shown in Fig. 1b. Although a smaller γ value is applied in TOPH, which means a stronger upper-level heating and weaker bottom cooling, its average heating profile is not changed much compared to CTRL. One possible reason is that the elevated heating in the upper troposphere in TOPH will stabilize the upper troposphere, thereby suppressing ensuing convection. Thus, on time average the difference is small. However, as will be seen later, the intraseasonal variability is considerably improved in TOPH compared to CTRL. There is a clear lower-level cooling in BOTC and BOTC2 with different amplitudes. All actual profiles of experiments show similar features to CTRL above around 700 hPa except TOPH2, which has a top-heavy shape peaking around 500 hPa. This is because, unlike other experiments, in TOPH2 cloud top is artificially set to 168 hPa even if the actual cloud top is at a much lower level.

Table 1.

Parameter setting for each experiment.

Parameter setting for each experiment.
Parameter setting for each experiment.

b. Data

The data used in this study include reanalysis data and model output from 1992 to 2001. The datasets include daily NOAA Interpolated Outgoing Longwave Radiation, High Resolution Infrared Radiation Sounder (HIRS) (Lee 2014), daily and six-hourly NCEP–DOE AMIP-II reanalysis zonal and meridional winds and specific humidity (Kanamitsu et al. 2002), daily TRMM precipitation product (Huffman et al. 2007), and daily GPCP product (Huffman et al. 2001). In the following analysis, we mainly use the anomaly fields, which are constructed by removing the annual mean from the original daily data and then applying a 20–80-day Lanczos bandpass filter. The analysis will focus on boreal winter and tropical area from 10°S to 10°N.

3. Results

a. Spectral analysis

To examine the intraseasonal variability and MJO simulation, we first conduct a spectral analysis using outgoing longwave radiation and precipitation anomalies. Wavenumber and frequency spectra for equatorial OLR anomalies from HIRS, CTRL, TOPH, and BOTC are shown in Fig. 2 for boreal winter. Observations show that there is a concentration of power at 30–80-day periods and zonal wavenumbers 1–3 for eastward propagation. There is also a secondary spectral peak of eastward movement at wavenumbers 2–4 and longer periods. There is more power in eastward propagation than westward propagation.

Fig. 2.

November–April wavenumber–frequency spectra of OLR anomalies averaged over 10°N–10°S for (a) NOAA satellites (HIRS), (b) CAM5.3 (CTRL), (c) TOPH, and (d) BOTC. Individual November–April spectra were calculated for each year and then averaged over all years of data. The climatological seasonal cycle and time mean for each November–April segment were removed before calculation of the spectra.

Fig. 2.

November–April wavenumber–frequency spectra of OLR anomalies averaged over 10°N–10°S for (a) NOAA satellites (HIRS), (b) CAM5.3 (CTRL), (c) TOPH, and (d) BOTC. Individual November–April spectra were calculated for each year and then averaged over all years of data. The climatological seasonal cycle and time mean for each November–April segment were removed before calculation of the spectra.

In model simulations, CTRL shows a symmetric pattern of eastward and westward spectra power with comparable amplitude at much longer periods (greater than 80 days). It peaks at zonal wavenumbers 1–3 in eastward propagation and wavenumbers 3 and 4 in westward propagation. The TOPH test has a similar power and pattern to the CTRL. For the BOTC experiment, it shows power in the proper wavenumber and frequency ranges but at significantly larger magnitude. Its eastward power is about 4 times that of westward power at intraseasonal time and spatial scales of MJO. Both observations and BOTC exhibit the most intense signal at the 30–80-day period and zonal wavenumbers 1–3 in the eastward direction. Figure 2 clearly shows that the MJO signal is indeed sensitive to the heating profile, and the one with lower-tropospheric cooling shows the closest resemblance to observations.

To further quantify the intraseasonal oscillation characteristics of each experiment, Fig. 3 shows wavenumber–frequency power spectra for the symmetric component of boreal winter equatorial rainfall using 6-hourly data. The GPCP observations show strong MJO and Kelvin wave signals. In the CTRL run, the Kelvin waves are simulated to some extent, but with weaker power. The MJO power is largely missing, consistent with what is seen in Fig. 2. When the top-heavy heating profile is used, the Kelvin waves become weaker and the MJO power is almost the same as in CTRL. In the lower-level cooling experiment, the improvement in MJO power is obvious. Within zonal wavenumbers 1–3 and 30–80-day frequency, BOTC has a comparable pattern and a slightly weaker magnitude in spectral power to the observations. However, the Kelvin waves are much weaker than in observations and other experiments.

Fig. 3.

Wavenumber–frequency spectra of the symmetric component of equatorial rainfall for (a) GPCP, (b) CTRL, (c) TOPH, and (d) BOTC. Spectra were computed for individual latitudes and then averaged over 30°N–30°S. Dispersion curves for the (n = −1) Kelvin, n = 1 equatorial Rossby (ER), n = 0 eastward inertiogravity (IG) modes in the shallow water equations. MJO is defined as the spectral components within zonal wavenumbers 1–3 and having time periods 30–80 days.

Fig. 3.

Wavenumber–frequency spectra of the symmetric component of equatorial rainfall for (a) GPCP, (b) CTRL, (c) TOPH, and (d) BOTC. Spectra were computed for individual latitudes and then averaged over 30°N–30°S. Dispersion curves for the (n = −1) Kelvin, n = 1 equatorial Rossby (ER), n = 0 eastward inertiogravity (IG) modes in the shallow water equations. MJO is defined as the spectral components within zonal wavenumbers 1–3 and having time periods 30–80 days.

b. Multivariate EOF

A composite life cycle of MJO-scale OLR and 850-hPa wind anomalies for boreal winter is shown in Fig. 4. Observations, TOPH, and BOTC show eastward propagation of convection (negative OLR anomalies) during the MJO life cycle from phase 1 in the Indian Ocean to phase 8 in the western Pacific, with similar propagation speed. There are easterlies ahead of the MJO convection and westerlies behind the MJO convection. The wind amplitude increases in the first half cycle and then becomes weaker in the observations; however, there are still strong winds even at phase 7 in the TOPH and BOTC. While TOPH also shows a clear eastward propagation of convection, the strength of the composite is weaker and the propagation is faster after phase 6 compared to the observation and BOTC. There are stronger winds than observations especially in the second half cycle in both TOPH and BOTC compared to the observations. CTRL shows some eastward propagation from the Indian Ocean across the Maritime Continent during the first several phases. However, the signals become much weaker and not well organized.

Fig. 4.

Composite November–April 20–100-day precipitation (color) and 850-hPa wind anomalies (vectors) as a function of MJO phase. (a) Reanalysis, NOAA Interpolated Outgoing Longwave Radiation, and NCEP–DOE AMIP-II reanalysis 850-hPa wind field; (b) CTRL; (c) TOPH; and (d) BOTC. The phases are determined from a two-dimensional phase diagram of the first two principal components (Wheeler and Hendon 2004). The phase and the number of data points that exceeded the threshold value for that phase are shown in the bottom right of each plot.

Fig. 4.

Composite November–April 20–100-day precipitation (color) and 850-hPa wind anomalies (vectors) as a function of MJO phase. (a) Reanalysis, NOAA Interpolated Outgoing Longwave Radiation, and NCEP–DOE AMIP-II reanalysis 850-hPa wind field; (b) CTRL; (c) TOPH; and (d) BOTC. The phases are determined from a two-dimensional phase diagram of the first two principal components (Wheeler and Hendon 2004). The phase and the number of data points that exceeded the threshold value for that phase are shown in the bottom right of each plot.

Figure 5 shows the Hovmöller diagram of 850-hPa zonal wind, OLR, and precipitation averaged over 10°S–10°N for reanalysis and the three simulations. In Fig. 5 (top), the shading and contours represent the composite OLR and 850-hPa zonal wind. In Fig. 5 (bottom), shading represents precipitation. It is seen that the wind anomalies propagate eastward over the entire equatorial belt, while the OLR and precipitation anomalies are limited to the west of the date line in the observations, TOPH, and BOTC experiments. In terms of phasing between OLR anomalies and 850-hPa wind anomalies, both the reanalyses and the two simulations show that convection is located in the westerly wind region, with easterly wind anomalies feeding into convection from ahead of the MJO as it propagates eastward. The MJO signals in precipitation are weaker in the TOPH than that in the observations and BOTC. In the CTRL, there are signals of eastward propagation from the Indian Ocean to the Maritime Continent. Afterward, the signals become incoherent. The zonal wind anomalies are in better agreement with the observations, and they propagate eastward along the global equatorial belt. Recently, when comparing MJO simulations by GCMs that participated in the MJO intercomparison project (Jiang et al. 2015), Wang et al. (2017) found that models that produce realistic MJO eastward propagation had upward motion west of the MJO convection and downward motion east of the MJO convection in the upper troposphere. By prescribing a stratiform-like heating anomaly in the rear of the MJO in an aquaplanet GCM, they found that the zonal asymmetry of vertical velocity anomaly about the MJO was well reproduced. They therefore suggest that stratiform heating is important to the simulation of propagating MJO. This is consistent with our results from the TOPH experiment, which shows improved MJO simulation compared to the CTRL run. Yet, BOTC further adds to the improvement when the cooling in the lower troposphere is included.

Fig. 5.

Phase–longitude Hovmöller plots of the (top) composite OLR (W m−2; shaded with a contour interval of 3 W m−2) and 850-hPa zonal wind (m s−1; contours with an interval of 0.5 m s−1) and (bottom) precipitation (mm day−1; contours with an interval of 0.5 mm day−1). All variables are averaged over 10°S–10°N for (a) reanalysis, NOAA Interpolated Outgoing Longwave Radiation, NCEP–DOE AMIP-II reanalysis 850-hPa wind field, and TRMM precipitation; (b) CTRL; (c) TOPH; and (d) BOTC.

Fig. 5.

Phase–longitude Hovmöller plots of the (top) composite OLR (W m−2; shaded with a contour interval of 3 W m−2) and 850-hPa zonal wind (m s−1; contours with an interval of 0.5 m s−1) and (bottom) precipitation (mm day−1; contours with an interval of 0.5 mm day−1). All variables are averaged over 10°S–10°N for (a) reanalysis, NOAA Interpolated Outgoing Longwave Radiation, NCEP–DOE AMIP-II reanalysis 850-hPa wind field, and TRMM precipitation; (b) CTRL; (c) TOPH; and (d) BOTC.

Using satellite observations, Del Genio et al. (2012) show that shallow and congestus clouds occur frequently ahead of the MJO peak phase, deep clouds occur near the peak, and upper-level anvils occur after the peak. To examine if the model simulations are able to capture this evolution of cloud population during MJO’s eastward propagation, we examine the cloud-top frequency in each MJO phase with output of cloud-top height at each time step. Figure 6 shows the frequency distribution of cloud-top height at each phase. The region of Maritime Continent (115°–125°E) is used here, since the MJO convection reaches the maximum in this region during its eastward propagation. Thus, the structural similarities and differences could be compared among the simulations and observations. For CTRL, the cloud-top heights are mainly in the range between 700 and 350 hPa through the whole MJO life cycle. There is a maximum of occurrence frequency for shallow convection with tops near 600 hPa at phase 2, and the cloud tops become higher up to 400 hPa by phase 3. The cloud-top heights generally increase with phases. In the TOPH experiment, the most frequently occurring cloud tops are lower compared to CTRL. However, the phase variation is more reasonable, with cloud top at higher altitudes at the MJO mature phase than other phases. The BOTC experiment shows a clear evolution of cloud-top height with MJO phases. There is mostly shallow convection at the first and second phases of MJO life cycle, and then there is a clear transition from shallow to deep convection, with the dominant cloud tops changing from below 500 hPa at phase 2 to 300–250 hPa by phase 5. Afterward, the cloud-top heights decrease with MJO phase to 500 hPa by phase 8. The transition from shallow to deep convection in the early phases of MJO in BOTC likely acts to provide the preconditioning needed in the MJO development.

Fig. 6.

Phase–height cross section of cloud-top frequency for an area (115°–125°E) over the Maritime Continent for (a) CTRL, (b) TOPH, and (c) BOTC.

Fig. 6.

Phase–height cross section of cloud-top frequency for an area (115°–125°E) over the Maritime Continent for (a) CTRL, (b) TOPH, and (c) BOTC.

To further examine the evolution of the moisture field during MJO eastward propagation, longitude–height plots of specific humidity anomalies are shown in Fig. 7. Observations show that there are moist anomalies in the Indian Ocean and dry anomalies in the western Pacific at phase 1. By phase 3, the moist anomalies have expanded to the west of 150°E. There is a clear westward tilt at the edge of the moist anomaly region, suggesting that moistening develops at the lower troposphere first. At phase 5, the positive moisture anomaly region has moved to the western Pacific, and the Indian Ocean is occupied by dry anomalies. A shallow layer of moist anomaly is clearly seen ahead of the deep moist layer. By phase 7, the positive moisture anomaly moves to the central and eastern Pacific, and the dry anomalies begin to move into the western Pacific. In both the TOPH and BOTC experiments, positive and negative moisture anomalies propagate eastward, much like in observations, including clear moistening in the lower troposphere ahead of the deep moist layer. The results from CTRL show similar features from Indian Ocean to the Maritime Continent but very little agreement with the observations afterward. In the CTRL, although there is eastward propagation of moisture anomalies at the beginning, the anomalies are not well organized compared to observations and the other two simulations after phase 3.

Fig. 7.

Longitude–height cross sections of specific humidity anomalies (g kg−1; contours with an interval of 0.05g kg−1) during MJO evolution at phases 1, 3, 5, and 7 for (a) NCEP–DOE AMIP-II reanalysis, (b) CTRL, (c) TOPH, and (d) BOTC.

Fig. 7.

Longitude–height cross sections of specific humidity anomalies (g kg−1; contours with an interval of 0.05g kg−1) during MJO evolution at phases 1, 3, 5, and 7 for (a) NCEP–DOE AMIP-II reanalysis, (b) CTRL, (c) TOPH, and (d) BOTC.

Further analysis of moisture budget using a method similar to Kiranmayi and Maloney (2011) is carried out to examine the role of vertical advection in the moistening of the lower troposphere before deep convection. The moisture budget equation is given by

 
formula

where q is specific humidity, V is the horizontal wind vector, ω is vertical pressure velocity, t is time, and Q2 is the apparent moisture sink associated with subgrid-scale condensation. The parentheses represent the vertical integral from 1000 to 700 hPa. Instead of integrating over the entire troposphere, we focus on the moisture source in the lower troposphere to identify the role of shallow convection. As shown in Fig. 8, vertical advection of moisture is much larger than horizontal advection in all three simulations. However, there are important differences among the simulations as well. In CTRL, positive vertical advection of moisture coincides with MJO convection as represented by precipitation. This is apparently due to vertical motion associated with deep convection. On the other hand, positive vertical moisture advection is well ahead of MJO deep convection in BOTC. This vertical motion is related to shallow convection. The TOPH simulation also has vertical moisture advection ahead of deep convection, but with shorter lead time. A recent study by Kiranmayi and Maloney (2011) through analysis of vertically integrated moist static energy budget shows that horizontal advection of moist static energy plays an important role in MJO onset in reanalysis data. As shown in Fig. 8, we further analyzed the moisture budget over the layer of 700–400 hPa (roughly the maximum moisture anomalies layer in Fig. 7) and found that horizontal advection of moisture is indeed an important contributor to moisture tendency in this layer ahead of MJO deep convection, whereas vertical advection coincides with MJO deep convection in all three simulations during the MJO development phases. However, the differences of vertical advection ahead of MJO in the lower troposphere between CTRL and the other two simulations are much larger than their differences in horizontal advection in the mid-to-upper levels. It indicates that the lack of contributions of vertical advection in the lower level to moisture tendency in CTRL is responsible for its poor MJO simulation in this study.

Fig. 8.

(first and third rows) Horizontal and (second and fourth rows) vertical advection of moisture vertically integrated from 1000 to 700 hPa and from 700 to 400 hPa (shaded) averaged from 10°S to 10°N for (a) CTRL, (b) TOPH, and (c) BOTC. Precipitation anomalies (mm day−1; contours with an interval of 0.5 mm day−1) are also plotted for reference.

Fig. 8.

(first and third rows) Horizontal and (second and fourth rows) vertical advection of moisture vertically integrated from 1000 to 700 hPa and from 700 to 400 hPa (shaded) averaged from 10°S to 10°N for (a) CTRL, (b) TOPH, and (c) BOTC. Precipitation anomalies (mm day−1; contours with an interval of 0.5 mm day−1) are also plotted for reference.

Figure 9 illustrates the longitude–height plots of total convective heating anomalies from both shallow and deep convection along the equator during MJO eastward propagation. The composite total precipitation in each phase is shown by the red line. BOTC shows that a deep layer of convective heating develops at phase 1 over the western Indian Ocean, corresponding to enhanced rainfall anomalies. Then the precipitation anomalies intensify while moving to the east. The heating shows a clear signal of eastward propagation and peaks in the lower to midtroposphere. Further comparison shows that positive heating anomalies are consistent with the moisture anomalies (Fig. 7), and there is westward tilting with altitude during the eastward propagation in total convective heating anomalies from both shallow and deep convection in the lower troposphere. It indicates the low-level heating associated with shallow convection leading deep convection. TOPH shows eastward propagation but much weaker signals compared to BOTC. There are also positive low-level heating anomalies ahead of deep convection in TOPH. Jiang et al. (2011) show similar features of heating anomalies in three different reanalysis datasets. For CTRL, during the first three phases, there are shallow convection anomalies ahead of deep convection, which is consistent with the relatively good simulation in the early phases in CTRL, as shown in Fig. 4. Afterward, shallow convective heating is largely absent ahead of MJO deep convection, and no well-established deep convection signals are seen.

Fig. 9.

Longitude–height cross sections of the total heating anomalies from both shallow and deep convection (K day−1; contours with an interval of 0.4 K day−1) and longitudinal distribution of precipitation anomalies (mm day−1; red line scales on right axis and black dashed line is zero line) during MJO evolution at phases 1, 3, 5, and 7 for (a) CTRL, (b) TOPH, and (c) BOTC.

Fig. 9.

Longitude–height cross sections of the total heating anomalies from both shallow and deep convection (K day−1; contours with an interval of 0.4 K day−1) and longitudinal distribution of precipitation anomalies (mm day−1; red line scales on right axis and black dashed line is zero line) during MJO evolution at phases 1, 3, 5, and 7 for (a) CTRL, (b) TOPH, and (c) BOTC.

To verify that the imposed stratiform-like heating, especially the lower-level cooling, is indeed directly responsible for the increased shallow convection, we designed a test experiment, in which the model configuration is the same as that of the CTRL run except that at each time step we make additional diagnostic calls. Specifically, at each time step after the calls to deep convection, shallow convection and cloud macrophysics subroutines, and the physical tendencies from these processes are saved, we go back to the deep convection heating profile, modify it as in BOTC, and update the physics state (to a temporary array) using the modified heating profile. Then, we proceed to call shallow convection and cloud macrophysics subroutines. These new physics tendencies are for diagnostic purposes and do not participate in the model integration for the next time step. The differences between the new and the original shallow convection and cloud macrophysics tendencies at each time step give the effect of increased lower-level cooling in the heating profile in BOTC on shallow convection and cloud condensation. Figure 10 shows the moisture tendency and shallow convective mass flux differences between the two types of the calls averaged over an area in the western Pacific (5°S–5°N, 135°–145°E) and a time period of each time step for two months. There is more shallow convective mass flux below 600 hPa, with a maximum near 920 hPa. Correspondingly, there is drying from the surface to 850 hPa and moistening from 850 to 600 hPa, a typical characteristic of shallow convection. In other words, there is much more shallow convection when upper-level heating and lower-level cooling are introduced to mimic stratiform condensational heating and evaporative cooling as compared to the CTRL run. The cloud macrophysics changes are relatively small, mostly acting to offset the shallow convective tendencies. This clearly demonstrates the role of stratiform-like heating and cooling in enhancing shallow convection.

Fig. 10.

Vertical profiles of differences in moisture tendency and shallow convective mass flux between the two calls averaged over an area of the western Pacific (5°S–5°N, 135°–145°E) and a time period of two months. The red and blue lines represent the moisture tendency differences after calling shallow convection and macrophysics modules, respectively. The black line represents the shallow convective mass flux difference between two processes after calling shallow convection module. See text for details.

Fig. 10.

Vertical profiles of differences in moisture tendency and shallow convective mass flux between the two calls averaged over an area of the western Pacific (5°S–5°N, 135°–145°E) and a time period of two months. The red and blue lines represent the moisture tendency differences after calling shallow convection and macrophysics modules, respectively. The black line represents the shallow convective mass flux difference between two processes after calling shallow convection module. See text for details.

c. Convection and low-level convergence

The analyses above indicate that the heating profile with stratiform-like heating simulates the intraseasonal variability well compared to observations, especially the one with lower-level cooling, and that there is a moisture preconditioning by shallow convection ahead of active MJO. The increased shallow convection as a direct result of the imposed mesoscale heating serves to precondition the lower-tropospheric moisture field for deep convection.

To further understand the roles of latent heating from convection and its interaction with PBL convergence in MJO development, the relationship between 850-hPa vertical velocity, a measure of low-level mass convergence (shaded), and 600–500-hPa average deep convective heating anomalies (contours) is shown in Fig. 11. In CTRL, the positions of 850-hPa vertical velocity coincide with those of midtropospheric convective heating in the western Pacific. The deep convective heating anomalies from the Indian Ocean to the Maritime Continent in the first three phases have little corresponding low-level mass convergence, indicating midlevel convection or absence of shallow convection. Afterward, the low-level convergence coincides with or slightly leads deep convection heating in the CTRL. On the other hand, the BOTC and TOPH show that low-level mass convergence is ahead of deep convection, as they propagate eastward throughout the life cycle of MJO. This is consistent with previous studies using global data assimilation products (Hendon and Salby 1994; Maloney and Hartmann 1998; Kiladis et al. 2005), which indicate that the boundary layer convergence leads the convection center of the MJO. Similarly, using reanalysis data, Hsu and Li (2012) suggest that the key element for eastward propagation of MJO is the boundary layer moisture asymmetry relative to the MJO convection. They attribute this moisture asymmetry mainly to moistening from advection of background moisture by the MJO ascending flow associated with the boundary layer convergence. They further argued that the free atmospheric wave dynamics induced by MJO convective heating accounts for most of the PBL convergence. With lifting from lower-level convergence, the potential instability may help trigger the shallow convection. The shallow convection transports moisture upward from the boundary layer into the lower troposphere (Figs. 8 and 10), leading to the onset of deep convection. The heating associated with MJO deep convection induces a baroclinic free atmospheric response to the east of the convective center, which may induce a convergent flow in the lower level. Thus, the convergence-generation process is associated with convective heating. Wu (2003) showed that lower-level heating from shallow convection could generate the boundary layer convergence. Zhang and Song (2009) also showed that shallow convection plays an important role in generating lower-level convergence through its heating to precondition deep convection in MJO. Compared to the corresponding phases of low-level convergence and deep convection heating in the CTRL, the leading convergence ahead of deep convection in TOPH and BOTC indicates that the modified heating indeed affects the generation of convergence. In addition, although both BOTC and TOPH show that low-level convergence anomalies lead deep convective heating, shallow convection is more ahead of the MJO convection in BOTC than in TOPH (Fig. 8). This larger phase lag indicates a more favorable condition provided by the enhanced shallow convection ahead of deep convection in BOTC because it gives more time for moistening the lower-tropospheric atmosphere.

Fig. 11.

Phase–longitude Hovmöller plots of low-level convergence anomalies (shaded) and 600–500-hPa average deep convective heating anomalies (contours) for (a) CTRL, (b) TOPH, and (c) BOTC. Positive values denote upward motion at 850 hPa or mass convergence below.

Fig. 11.

Phase–longitude Hovmöller plots of low-level convergence anomalies (shaded) and 600–500-hPa average deep convective heating anomalies (contours) for (a) CTRL, (b) TOPH, and (c) BOTC. Positive values denote upward motion at 850 hPa or mass convergence below.

4. Additional sensitivity tests

As shown in section 3, the simulated MJOs are more realistic in TOPH and BOTC. This suggests that including a stratiform heating and cooling structure can indeed improve the MJO simulation, and evaporative cooling in the lower troposphere might be more important compared to top-heavy heating. In this section, we examine how sensitive these simulations are to further modifications. As stated in section 2a, climatologically cloud tops in CAM5.3 are generally lower than observed. Will this affect the MJO simulation? This will be tested in the TOPH2 experiment. As we showed, increasing the lower-level cooling (e.g., cf. TOPH and BOTC) results in a relatively better MJO simulation. What if the lower-level cooling is further increased? This is tested in the BOTC2 experiment. Figure 12 shows the wavenumber–frequency power spectra of rainfall for experiments TOPH2 and BOTC2. In TOPH2, the Kelvin wave is greatly enhanced. In fact, it is too strong compared to the reanalysis data (Fig. 3). The MJO simulation is deteriorated compared to TOPH. It is not clear to us why increasing the convective top height results in an enhanced Kelvin wave power; it warrants further study in the future. The result of BOTC2 is similar to the BOTC experiment in that there is clear MJO power, although both the eastward- and westward-propagating signals are weaker than in BOTC. Interestingly, the Kelvin wave power becomes weaker, again for reasons unknown to us. Figure 13 shows the life cycle of composite OLR and 850-hPa wind field in boreal winter. Compared to TOPH, TOPH2 shows a much weaker MJO signal although there is some indication of eastward propagation, consistent with the power spectrum in Fig. 12. BOTC2 is similar to BOTC except the propagation speed is slightly faster after phase 6 and the magnitude of the anomalies is larger. These two additional experiments suggest that top-heavy heating may play a less important role than previously thought. On the other hand, lower-level cooling is more essential to a realistic MJO simulation.

Fig. 12.

As in Fig. 3, but for (a) TOPH2 and (b) BOTC2.

Fig. 12.

As in Fig. 3, but for (a) TOPH2 and (b) BOTC2.

Fig. 13.

As in Fig. 4, but for (a) TOPH2 and (b) BOTC2.

Fig. 13.

As in Fig. 4, but for (a) TOPH2 and (b) BOTC2.

5. Summary and conclusions

This study investigates the role of diabatic heating in simulating MJO in the NCAR CAM5.3. By modifying the heating profile produced by the ZM deep convection scheme, we investigate the sensitivity of MJO simulation to different shapes of heating profile from top-heavy heating to bottom-heavy cooling to mimic the stratiform heating and cooling using a method similar to that of Moncrieff and Liu (2006). Our results suggest that the mesoscale stratiform heating is important to MJO simulation, especially its cooling component in the lower levels. By artificially adding stratiform-like heating and cooling in the experiments, it gives better simulations of MJO that are in good agreement with observations, including intraseasonal spectral power, realistic westward-tilting vertical structure, and eastward propagation with proper speed. Its composite MJO signals for OLR, 850-hPa wind field, and precipitation are all comparable to observations and reanalysis. There are positive moisture anomalies ahead of the main MJO convection, corresponding to shallow convection, indicating that enhanced shallow convection plays an important role in preconditioning the atmosphere in the development of MJO intraseasonal variability. Further comparison shows that lower-level cooling contributes more to improving the MJO simulation. Further sensitivity experiments show that the improvement of MJO simulation is not too sensitive qualitatively to the strength of lower-level cooling, although quantitative differences are noticeable. In contrast, both the CTRL simulation and the TOPH2 simulation fail to produce realistic MJO.

It is further shown that the imposed stratiform-like heating by modifying convective heating from the ZM scheme is directly responsible for the increased shallow convection. The increased convective mass flux from shallow convection produces more drying in the boundary layer and more moistening in the lower troposphere above it. By carrying out a moisture budget analysis in the lower troposphere from 1000 to 700 hPa, it is shown that advection of moisture by vertical motion associated with shallow convection contributes to the moisture preconditioning ahead of the MJO convection in both TOPH and BOTC. Compared to CTRL, the role of shallow convection in improving the MJO simulation is significant. The TOPH and BOTC experiment also show boundary layer convergence ahead of the deep convection, consistent with the observational and modeling work of Hsu and Li (2012) and Hsu et al. (2014). In addition, Anber et al. (2016) suggest that with the strong wind shear and large surface flux during active MJO, cloud radiative interaction has a significant impact on MJO in the presence of a substantial layer of high anvil cloud and slightly stronger descending motion in the lower troposphere, which resembles the effect of upper-level heating and lower-level cooling in this study. It indicates that the heating profile could feed back on the large-scale wind field. Such interactions provide a possible physical explanation that the convergence in the midtroposphere caused by cloud interaction may contribute to the enhanced shallow convections after adding stratiform-like heating.

Our results suggest that incorporating a mesoscale stratiform heating structure can improve the MJO simulation in the NCAR CAM5.3. Particularly, the lower-level cooling plays a more important role than the upper-level heating in producing intraseasonal variability. While this study demonstrates the improvement of MJO simulation in the NCAR CAM5.3 by adding mesoscale stratiform heating and cooling in the deep convection, it will be interesting to see if other models behave similarly. In this regard, the work of Wang et al. (2017) is interesting. By analyzing MJO simulations from more than 20 current GCMs, they suggest that the representation of diabatic heating structure, especially the upper-level stratiform-like heating, is critical to realistic simulations of MJO. Whereas both our study and the work of Wang et al. (2017) emphasize the role of stratiform heating structure in MJO simulation, each offers a different perspective of the relative roles of upper-level heating and lower-level cooling. The experiments in this study are relatively extreme in that the heating profile is modified whenever there is deep convection with cloud top above the 700-hPa level. Deep convection can exist without producing mesoscale heating and cooling. The real challenge is how to parameterize convective organization that is linked to mesoscale heating and cooling. Mapes and Neale (2011) have attempted to parameterize mesoscale organization, but its effect on MJO simulation has yet to be seen.

Acknowledgments

This material is based upon work supported by the China Meteorological Administration Special Public Welfare Research Fund GYHY201406007, the U.S. National Science Foundation Grant AGS-1549259, and the U.S. Department of Energy, Office of Science, Biological and Environmental Research Program (BER) Award DE-SC0016504. The authors thank three anonymous reviewers for their constructive and insightful comments, which helped improve the quality of the manuscript. The abstract of this work was submitted to the 2017 EGU General Assembly.

REFERENCES

REFERENCES
Anber
,
U.
,
S.
Wang
,
A. H.
Sobel
, and
D. M.
Frierson
,
2016
:
Response of atmospheric convection to vertical wind shear: Cloud-system-resolving simulations with parameterized large-scale circulation. Part II: Effect of interactive radiation
.
J. Atmos. Sci.
,
73
,
199
299
, doi:.
Barnes
,
H. C.
,
M. D.
Zuluaga
, and
R. A.
Houze
Jr.
,
2015
:
Latent heating characteristics of the MJO computed from TRMM observations
.
J. Geophys. Res. Atmos.
,
120
,
1322
1334
, doi:.
Del Genio
,
A. D.
,
Y.
Chen
,
D.
Kim
, and
M. S.
Yao
,
2012
:
The MJO transition from shallow to deep convection in CloudSat/CALIPSO data and GISS GCM simulations
.
J. Climate
,
25
,
3755
3770
, doi:.
Donner
,
L. J.
,
1993
:
A cumulus parameterization including mass fluxes, vertical momentum dynamics, and mesoscale effects
.
J. Atmos. Sci.
,
50
,
889
906
, doi:.
Fu
,
X.
, and
B.
Wang
,
2009
:
Critical roles of the stratiform rainfall in sustaining the Madden–Julian oscillation: GCM experiments
.
J. Climate
,
22
,
3939
3959
, doi:.
Hartmann
,
D. L.
,
H. H.
Hendon
, and
R. A.
Houze
Jr.
,
1984
:
Some implications of the mesoscale circulations in tropical cloud clusters for large-scale dynamics and climate
.
J. Atmos. Sci.
,
41
,
113
121
, doi:.
Hendon
,
H. H.
, and
M. L.
Salby
,
1994
:
The life cycle of the Madden–Julian oscillation
.
J. Atmos. Sci.
,
51
,
2225
2237
, doi:.
Houze
,
R. A.
,
1989
:
Observed structure of mesoscale convective systems and implications for large‐scale heating
.
Quart. J. Roy. Meteor. Soc.
,
115
,
425
461
, doi:.
Hsu
,
P. C.
, and
T.
Li
,
2012
:
Role of the boundary layer moisture asymmetry in causing the eastward propagation of the Madden–Julian oscillation
.
J. Climate
,
25
,
4914
4931
, doi:.
Hsu
,
P. C.
,
T.
Li
, and
H.
Murakami
,
2014
:
Moisture asymmetry and MJO eastward propagation in an aquaplanet general circulation model
.
J. Climate
,
27
,
8747
8760
, doi:.
Huffman
,
G. J.
,
R. F.
Adler
,
M. M.
Morrissey
,
D. T.
Bolvin
,
S.
Curtis
,
R.
Joyce
,
B.
McGavock
, and
J.
Susskind
,
2001
:
Global precipitation at one-degree daily resolution from multisatellite observations
.
J. Hydrometeor.
,
2
,
36
50
, doi:.
Huffman
,
G. J.
,
D. T.
Bolvin
,
E. J.
Nelkin
,
D. B.
Wolff
,
R. F.
Adler
,
G.
Gu
, and
E. F.
Stocker
,
2007
:
The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales
.
J. Hydrometeor.
,
8
,
38
55
, doi:.
Jiang
,
X.
, and Coauthors
,
2009
:
Vertical heating structures associated with the MJO as characterized by TRMM estimates, ECMWF reanalyses, and forecasts: A case study during 1998/99 winter
.
J. Climate
,
22
,
6001
6020
, doi:.
Jiang
,
X.
, and Coauthors
,
2011
:
Vertical diabatic heating structure of the MJO: Intercomparison between recent reanalyses and TRMM estimates
.
Mon. Wea. Rev.
,
139
,
3208
3223
, doi:.
Jiang
,
X.
, and Coauthors
,
2015
:
Vertical structure and physical processes of the Madden–Julian oscillation: Exploring key model physics in climate simulations
.
J. Geophys. Res. Atmos.
,
120
,
4718
4748
, doi:.
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
, doi:.
Kiladis
,
G. N.
,
K. H.
Straub
, and
P. T.
Haertel
,
2005
:
Zonal and vertical structure of the Madden–Julian oscillation
.
J. Atmos. Sci.
,
62
,
2790
2809
, doi:.
Kiranmayi
,
L.
, and
E. D.
Maloney
,
2011
:
Intraseasonal moist static energy budget in reanalysis data
.
J. Geophys. Res.
,
116
,
D21117
, doi:.
Lappen
,
C. L.
, and
C.
Schumacher
,
2012
:
Heating in the tropical atmosphere: What level of detail is critical for accurate MJO simulations in GCMs?
Climate Dyn.
,
39
,
2547
2568
, doi:.
Lau
,
K. M.
, and
L.
Peng
,
1987
:
Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory
.
J. Atmos. Sci.
,
44
,
950
972
, doi:.
Lee
,
H. T.
,
2014
: Climate algorithm theoretical basis document (C-ATBD): Outgoing longwave radiation (OLR)—Daily. NOAA Climate Data Record Program Rep. CDRP-ATBD-0526, 46 pp.
Li
,
C.
,
1983
:
Convection condensation heating and unstable modes in the atmosphere (in Chinese)
.
Chin. J. Atmos. Sci.
,
7
,
260
268
.
Li
,
C.
,
X.
Jia
,
J.
Ling
,
W.
Zhou
, and
C.
Zhang
,
2009
:
Sensitivity of MJO simulations to diabatic heating profiles
.
Climate Dyn.
,
32
,
167
187
, doi:.
Lin
,
J.
,
B.
Mapes
,
M.
Zhang
, and
M.
Newman
,
2004
:
Stratiform precipitation, vertical heating profiles, and the Madden–Julian oscillation
.
J. Atmos. Sci.
,
61
,
296
309
, doi:.
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
, doi:.
Madden
,
R. A.
, and
P. 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:.
Maloney
,
E. D.
, and
D. L.
Hartmann
,
1998
:
Frictional moisture convergence in a composite life cycle of the Madden–Julian oscillation
.
J. Climate
,
11
,
2387
2403
, doi:.
Mapes
,
B. E.
,
2000
:
Convective inhibition, subgrid-scale triggering energy, and stratiform instability in a toy tropical wave model
.
J. Atmos. Sci.
,
57
,
1515
1535
, doi:.
Mapes
,
B. E.
, and
R. B.
Neale
,
2011
:
Parameterizing convective organization to escape the entrainment dilemma
.
J. Adv. Model. Earth Syst.
,
3
,
M06004
, doi:.
Moncrieff
,
M. W.
, and
C.
Liu
,
2006
:
Representing convective organization in prediction models by a hybrid strategy
.
J. Atmos. Sci.
,
63
,
3404
3420
, doi:.
Mu
,
M.
, and
G. J.
Zhang
,
2006
:
Energetics of Madden–Julian oscillations in the National Center for Atmospheric Research Community Atmosphere Model version 3 (NCAR CAM3)
.
J. Geophys. Res.
,
111
,
D24112
, doi:.
Mu
,
M.
, and
G. J.
Zhang
,
2008
:
Energetics of Madden Julian oscillations in the NCAR CAM3: A composite view
.
J. Geophys. Res.
,
113
,
D05108
, doi:.
Neale
,
R. B.
,
J. H.
Richter
, and
M.
Jochum
,
2008
:
The impact of convection on ENSO: From a delayed oscillator to a series of events
.
J. Climate
,
21
,
5904
5924
, doi:.
Neale
,
R. B.
,
C. C.
Chen
,
A.
Gettelman
,
P. H.
Lauritzen
,
S.
Park
,
D. L.
Williamson
, and
D.
Marsh
,
2010
: Description of the NCAR Community Atmosphere Model (CAM 5.0). NCAR Tech. Note NCAR/TN-486+STR, 268 pp. [Available online at www.cesm.ucar.edu/models/cesm1.1/cam/docs/description/cam5_desc.pdf.]
Park
,
S.
, and
C. S.
Bretherton
,
2009
:
The University of Washington shallow convection and moist turbulence schemes and their impact on climate simulations with the Community Atmosphere Model
.
J. Climate
,
22
,
3449
3469
, doi:.
Schumacher
,
C.
,
R. A.
Houze
Jr.
, and
I.
Kraucunas
,
2004
:
The tropical dynamical response to latent heating estimates derived from the TRMM Precipitation Radar
.
J. Atmos. Sci.
,
61
,
1341
1358
, doi:.
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
, doi:.
Wang
,
L.
,
T.
Li
,
E. D.
Maloney
, and
B.
Wang
,
2017
:
Fundamental causes of propagating and nonpropagating MJOs in MJOTF/GASS models
.
J. Climate
,
30
,
3743
3769
, doi:.
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
, doi:.
Wing
,
A. A.
, and
K. A.
Emanuel
,
2014
:
Physical mechanisms controlling self-aggregation of convection in idealized numerical modeling simulation
.
J. Adv. Model. Earth Syst.
,
6
,
59
74
, doi:.
Wu
,
Z.
,
2003
:
A shallow CISK, deep equilibrium mechanism for the interaction between large-scale convection and large-scale circulations in the tropics
.
J. Atmos. Sci.
,
60
,
377
392
, doi:.
Zhang
,
C.
,
2005
:
Madden–Julian oscillation
.
Rev. Geophys.
,
43
,
RG2003
, doi:.
Zhang
,
G. J.
, and
N. A.
McFarlane
,
1995
:
Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model
.
Atmos.–Ocean
,
33
,
407
446
, doi:.
Zhang
,
G. J.
, and
M.
Mu
,
2005
:
Effects of modifications to the Zhang–McFarlane convection parameterization on the simulation of the tropical precipitation in the National Center for Atmospheric Research Community Climate Model, version 3
.
J. Geophys. Res.
,
110
,
D09109
, doi:.
Zhang
,
G. J.
, and
X.
Song
,
2009
:
Interaction of deep and shallow convection is key to Madden–Julian oscillation simulation
.
Geophys. Res. Lett.
,
36
,
L09708
, doi:.

Footnotes

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