1. Introduction
The Madden–Julian oscillation (MJO) is a dominant mode of intraseasonal variability in the tropics (Madden and Julian 1971). Decades of observational studies have documented the evolution of the MJO and its characteristics (Zhang 2005). In a review paper, Zhang (2005) summarized the current knowledge on the MJO and the challenge in understanding its mechanisms, simulation, and prediction. While noticeable progress has been made in documenting the multiscale structures of the MJO and associated cloud population, we are still faced with many obstacles in understanding their vertical structures. In the model simulation front, he noted that few models can simulate the observed structure of MJOs, and one of the possible causes is the convection parameterization and associated vertical heating distribution [see Zhang (2005) for details]. The MJO is a well-coupled system between convection and large-scale atmospheric circulation in which convective anomalies propagate eastward at an average speed of 5 m s−1 over the tropical Indian and western Pacific Oceans, accompanied with an anomalous large-scale circulation. The typical lifetime of the MJO ranges from 20 to 100 days. Several observational studies show a zonally asymmetric vertical structure of moisture, temperature, and mass convergence with low-level moistening, heating, and convergence leading the convection center of the MJO, resulting in a westward tilt (Maloney and Hartmann 1998; Kemball-Cook and Weare 2001; Hsu and Li 2012).
The MJO simulation is a challenging task for climate models. A great deal of improvement (e.g., greater eastward propagating power with reasonable zonal scales and periods) has been made since the 1980s by improving the cumulus parameterization (Maloney and Hartmann 2001), increasing the vertical resolution (Inness et al. 2001), and including feedback from intraseasonal sea surface temperature perturbation (Waliser et al. 1999; Benedict and Randall 2011). Nevertheless, there are still serious deficiencies in simulated MJOs, including more power at higher frequencies, underestimation of the strength of MJO, and failure to capture the seasonality of MJO events (Zhang and Mu 2005a; Zhang 2005; Kim et al. 2009). Observational and modeling studies have demonstrated that shallow convection has a nonnegligible heating and moistening effect on preconditioning the atmosphere for deep convection, and the interaction between deep and shallow convection is crucial to MJO simulation (Benedict and Randall 2007; Li et al. 2009). Zhang and Song (2009) found that the MJO signals were very weak over the Indian and the western Pacific Oceans when shallow convection in the National Center for Atmospheric Research (NCAR) Community Atmosphere Model, version 3.0 (CAM3.0), was suppressed in the tropical belt. Their study found that shallow convection served to precondition the lower troposphere by enhancing low-level mass convergence ahead of deep convection.
The mechanisms for the zonally asymmetric vertical structure with respect to the convective center of the MJO are unclear. One school of thought maintains that the low-level moistening prior to deep convection is regulated by frictional convergence (Wang 1988; Salby and Hendon 1994). The convergence in the boundary layer due to friction in advance of deep convection fosters low-level growth of moisture anomalies and makes the environment favorable for convection (Maloney and Hartmann 1998). Tian et al. (2006) found that the results of satellite data are consistent with the frictional Kelvin–Rossby wave/conditional instability of the second kind (CISK) theory in which the surface frictional mass convergence associated with equatorial Kelvin waves results in low-level moistening and heating prior to deep convection.
The recharge–discharge mechanism in MJO dynamics has been proposed to explain the time scale of the MJO through the interaction between convection and environmental humidity (Bladé and Hartmann 1993). It is characterized by a buildup of low-level moist static energy (MSE) before the occurrence of MJO deep convection and a discharge of convective energy during and after MJO deep convection (Maloney 2009; Del Genio et al. 2011). Since no single theory is able to explain all aspects of the MJO, several studies suggested that the recharge–discharge mechanism might be partially responsible for setting the period, and frictional-convergence feedback controls the energetics of the oscillation (Kemball-Cook and Weare 2001; Benedict and Randall 2007).
The significance of free-tropospheric water vapor for convection has also been recognized for both tropical convection in general (Bretherton et al. 2004; Peters and Neelin 2006) and the MJO in particular (Grabowski and Moncrieff 2004). By relating precipitation to column moisture, Raymond and colleagues (Raymond and Fuchs 2007, 2009; Raymond et al. 2009) identified a moisture mode that shows many of the MJO characteristics. In the moisture mode, a fundamental parameter is the gross moist stability (GMS), defined as proportional to the net export of moist static energy out of an atmospheric column (Raymond and Fuchs 2007). The gist of the moisture mode is that convection occurs when the GMS of the atmospheric column is negative. Convection leads to a net import of moist static energy into the atmospheric column (thus negative GMS), which leads to further enhanced convection resulting in an amplification of the initial perturbation (Fuchs and Raymond 2005; Raymond and Fuchs 2007). Through modeling analysis, Raymond and Fuchs (2007, 2009) and Raymond et al. (2009) argue that the MJO is driven at least partly by moisture mode instability.
As GMS involves column-integrated MSE, several studies have explored the vertically integrated MSE budget to understand the MJO evolution in models (Maloney 2009; Andersen and Kuang 2012) and reanalysis data (Kiranmayi and Maloney 2011). The advantage of using a vertically integrated MSE budget is that it can be directly related to top-of-atmosphere (and surface) radiative fluxes and surface turbulent fluxes. In this paper, we perform MSE and moisture budget analysis on the MJO evolution to evaluate the role of shallow convection in MJO simulation. This work differs from previous studies in the following ways. First, while most MSE budget analyses associated with the MJO focus on its column-integrated values, the vertical structure is also of vital importance to understand the processes in the MJO evolution (Benedict and Randall 2007, 2009). This work emphasizes the changes in vertical structure during MJO evolution. Second, while Zhang and Song (2009) have explored the effect of shallow convection on MJO simulation by examining the atmospheric fields (e.g., moisture, winds, mass convergence, etc.), this work extends their work by conducting a MSE budget analysis.
The remainder of the paper is organized as follows. In sections 2 and 3, an overview of the model, experiments, data, and method used in this study is given. Section 4 presents the main results, including the climatology, MJO simulation, MSE, and moisture budgets and the contribution from horizontal advection. Section 5 summarizes the paper.
2. Model simulations and data
The NCAR CAM3.0 (Collins et al. 2006) is used in this study. CAM3.0 has a horizontal resolution of T42 and a vertical resolution of 26 levels. The parameterization of large-scale cloud condensation and evaporation is described in Zhang et al. (2003) and Rasch and Kristjánsson (1998). Deep convection is parameterized using a revised Zhang–McFarlane scheme (Zhang 2002; Zhang and Mu 2005a,b). Compared with the original Zhang–MacFarlane scheme (Zhang and McFarlane 1995), three modifications are made. First, a new closure based on a quasi-equilibrium between convection and large-scale processes in the free troposphere is used. Second, the relative humidity of 80% is used as a convection trigger. Third, the revised scheme allows convection to originate above the planetary boundary layer (PBL) top. Shallow convection is parameterized by the Hack scheme (Hack 1994).
As in Zhang and Song (2009), using CAM3.0 we conducted two experiments to understand the role of deep and shallow convection in MJO simulation: the standard run (CTL) and the experiment run (NSC), which is the same as the CTL except that shallow convection is turned off below 700 hPa between 20°S and 20°N. Since shallow convection in the trade wind regime of the subtropical regions is an important climate feature (Stevens 2007), suppressing it everywhere would lead to too much distortion of the simulated mean climate. Thus we disable it only in the tropical belt. The instability below 700 hPa as a result of this modification is partly removed by PBL diffusion and partly built up for deep convection. For both simulations, the model integration is for 10 years starting from 1 September 1979 forced with observed monthly sea surface temperature (SST).
The Advanced Very High Resolution Radiometer (AVHRR) outgoing longwave radiation (OLR) (Liebmann and Smith 1996) is used as a proxy for convective activity in this study. The atmospheric fields are from the daily product of the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) (Simmons and Gibson 2000). We also examined the Atmospheric Model Intercomparison Project II (AMIP-II) reanalysis from the National Centers for Environmental Prediction (NCEP)–NCAR (herein NCEP2) (Kanamitsu et al. 2002). Several previous studies (Tian et al. 2010; Benedict and Randall 2007; Kim et al. 2009) have suggested that the NCEP2 data may have shortcomings in describing the vertical structures of the MJO thermodynamic fields. We also found that the NCEP2 has weaker MJO signals. Therefore, we will use the ERA-40 product in this paper. It should be noted that any reanalysis data is a blend of observations and model physics parameterization of the analysis/forecast system and thus should not be viewed as ground truth. This point should be taken into account when comparing the simulation results with the reanalysis.
3. Method
4. Results
In this section, we will first present the mean features of the model results. Then the simulated MJOs in the CTL and NSC (no shallow convection) runs are compared with those in the reanalysis data. Based on these results, we further perform detailed analysis on the MSE and moisture budget of the composite MJO over its life cycle.
a. General features
A realistic mean state of the atmosphere has been considered important to MJO simulation in GCMs (Kim et al. 2009). Previous studies showed that the strong low-level easterly wind in the tropical western Pacific is improved by using the revised Zhang and MacFarlane scheme (Zhang and Mu 2005a,b; Mu and Zhang 2006). Figure 1 shows the mean 850-hPa winds and OLR during boreal winter in the reanalysis data and CTL and NSC runs, respectively. In the reanalysis data (Fig. 1a), there is strong convection centered over the eastern Indian Ocean, Maritime Continent, and tropical western Pacific warm pool region associated with wind convergence at 850 hPa. The westerly wind extends to the western Pacific from the Indian Ocean between the equator and 10°S. The overall geographic patterns of the 850-hPa winds and OLR are reasonably captured in both simulations, although there are some quantitative differences between the model simulations and the reanalysis data (Figs. 1b,c). For instance, in both the CTL and NSC runs convection over the eastern Indian Ocean and the western Pacific warm pool region is weaker than that in the reanalysis data, especially in Maritime Continent, and the bias is more pronounced in the NSC run. It is noted that the westerly wind in the eastern Indian Ocean is absent in the CTL run while is reproduced in the NSC run. The impact of this will be discussed later in this section.
Mean state of OLR (color; W m−2) and winds at 850 hPa (vectors; m s−1) during boreal winter in the (a) ERA-40, (b) CTL, and (c) NSC runs.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
The differences in the mean specific humidity between the CTL simulation and ERA-40 and between NSC and CTL are shown in Figs. 2a and 2b. The CTL run is more moist near the surface, but drier in the free troposphere than the reanalysis from the eastern Indian Ocean to the western Pacific warm pool (Fig. 2a). These biases are amplified when shallow convection in the tropics is suppressed (Fig. 2b). There are moist anomalies in the planetary boundary layer and dry anomalies between 600 and 850 hPa in the NSC compared with that in the CTL. Together with the weaker vertical velocity in the NSC run except near 120°E (Fig. 2d), this directly reflects the fact that, without shallow convection, moisture is accumulated in the PBL and the vertical transport of moisture to the lower troposphere is ineffective. Correspondingly, there is more cloud in the PBL and less cloud in the lower troposphere in the NSC run than in the CTL (Fig. 2c). Because of the lack of shallow convection, deep convection is more active/frequent, leading to more high-level clouds.
The difference of (a),(b) specific humidity (g kg−1), (c) cloud fraction (%), and (d) vertical velocity (1 × 10−2 Pa s−1) in the mean state averaged over 10°S–10°N during boreal winter between (top) the CTL run and ERA-40 and (all others) the NSC and CTL runs.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
The observed and simulated phase–longitude cross sections of OLR and zonal wind at 850 hPa on MJO scale are presented in Fig. 3. In the reanalysis data (Fig. 3a), negative OLR anomalies are seen over the Indian Ocean at phase 1, indicating new MJO convection. After phase 2, convection in the Indian Ocean intensifies, accompanied by westerly wind anomalies. At phase 5, the MJO convection moves to the eastern Indian Ocean, where the strong negative OLR anomalies are seen near 80°E and 850-hPa wind convergence leads the MJO convection. During the subsequent phases, the convection gradually becomes weaker and moves eastward to the date line.
Phase–longitude cross section of OLR (color; W m−2) and zonal wind at 850 hPa (contours; m s−1) on the MJO scale averaged over 10°S–10°N in the (a) ERA-40, (b) CTL, and (c) NSC runs.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
The CTL run shows a good performance in reproducing the MJO convection, its eastward propagation, and the coupled structure between OLR and 850-hPa winds during the composite MJO life cycle (Fig. 3b). The zonal wind anomalies are also comparable to those in the ERA-40. However, the OLR amplitude is considerably weaker, especially in the Indian Ocean, and the propagation is faster than in the reanalysis data. In the NSC run (Fig. 3c), the MJO signals in OLR and zonal wind are much weaker than in the CTL run although they also show eastward propagation. As mentioned before, in the CTL run the climatological easterly wind at 850 hPa prevails in the eastern Indian Ocean, contrary to the westerly wind in the reanalysis data (Figs. 1a,b). However, the anomalies in zonal wind on MJO scale are reproduced reasonably well (Fig. 3b). On the other hand, for the NSC run, in spite of a better simulation of the mean zonal wind at 850 hPa (Fig. 1c), the anomalies on the MJO scale in the eastern Indian Ocean are much weaker than in the reanalysis data (Fig. 3c). This implies that the mean zonal wind, while important to MJO simulation, is not a deciding factor for MJO simulation. But shallow convection is.
b. MSE analysis
In this section, we analyze the MSE evolution during the MJO life cycle. Because strong MJO convection at phase 5 occurs near 80°E in the reanalysis data and 120°E in both the simulations (Figs. 3a–c), our budget analyses will focus on these regions: 80°–90°E in the ERA-40 and 120°–130°E in both the runs over the equatorial belt from 10°S to 10°N.
Figure 4 shows the phase–height cross sections of MSE and composite OLR anomalies. In the reanalysis data, at the beginning of MJO life cycle with the maximum OLR anomalies (Fig. 4d), negative MSE anomalies are seen below 200 hPa and positive anomalies above (Fig. 4a), indicating that the atmospheric column is more stable than the mean atmospheric state. After phase 1, this stable vertical structure gradually weakens. Positive MSE anomalies start to develop at 850 hPa and continue to build up with height. At phase 3 when convection occurs, as indicated by the OLR anomalies in Fig. 4d, positive MSE anomalies occupy the entire troposphere below 200 hPa. At phase 4, just before the maximum convection phase, MSE anomalies peak at 700 hPa. At this point the atmosphere is the most unstable. By phase 5 when the strongest convection occurs (Fig. 4d), MSE anomalies begin to weaken due to consumption of convective instability by convection, with negative anomalies developing at 850 hPa. Starting at phase 6 and onward, negative MSE anomalies gradually build up, reaching a maximum at phase 8 and composite MJO life cycle returns to the convection-suppressed phases. This MJO evolution is consistent with the recharge–discharge paradigm.
Height–phase cross section of (a)–(c) MSE (K) and (d) OLR (W m−2) on the MJO scale averaged over 10°S–10°N, 80°–90°E in (a) ERA-40, and over 10°S–10°N, 120°–130°E in (b) CTL and (c) NSC runs. The regions above 90% confidence level are stippled.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
The CTL run captures the overall MSE evolution well (Fig. 4b). It simulates the increasingly positive MSE anomalies at phases 1 and 2, with the anomalies first appearing near the surface and gradually moving upward. Similar to the reanalysis data, positive MSE anomalies occupy the entire troposphere below 200 hPa after phase 3 and the maximum MSE anomalies occur at phase 4 at 700 hPa. At the MJO mature phase (phase 5), MSE anomalies peak at 500 hPa and, at the same time, negative anomalies begin to appear near the surface and intensify afterward, again in qualitative agreement with the ERA-40 data. However, compared to ERA-40, the magnitude of the anomalies is significantly smaller, which is consistent with the weaker convection in Fig. 3b.
Without shallow convection (Fig. 4c), the NSC run has some different structure of MSE anomalies. The development of positive MSE anomalies is slower compared with the CTL run; for example, positive MSE anomalies do not occupy the entire troposphere below 200 hPa until after phase 4, corresponding to the negative OLR anomalies (Fig. 4d). The maximum MSE anomalies remain confined to below 850 hPa at phase 4, and are much weaker than in the CTL run. The decay of positive MSE anomalies is slower and still positive in the whole troposphere below 200 hPa at phase 6.
Having examined the evolution of MSE anomalies during the composite MJO life cycle, we next analyze the contributions from each MSE budget term to these anomalies. Figure 5 shows the temporal change and horizontal and vertical advection of MSE as functions of phase in the composite MJO life cycle. The 
Height–phase cross section of each term (K day−1) for the MSE budget averaged over (top) 10°S–10°N, 80°–90°E in ERA-40 and over 10°S–10°N, 120°–130°E in the (middle) CTL and (bottom) NSC runs, showing (a),(d),(g) total tendency, (b),(e),(h) horizontal advection, and (c),(f),(i) vertical advection. The regions above 90% confidence level are stippled.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
The CTL run (Figs. 5d–f) simulates all three budget terms and their evolution well, except that the horizontal advection below 800 hPa is positive during phases 6–7 and negative during phases 1–3 (Fig. 5e) and the amplitudes are larger compared with that in the ERA-40 (Figs. 5a–c).
Without shallow convection, the evolution pattern of the MSE budget shows some different features. The positive total tendency is weaker and the peak at the early stage of the MJO (e.g., phase 2) is located at 850 hPa (Fig. 5g), which is largely due to contributions from horizontal advection (Fig. 5h). The negative advection during phases 4–6 below 800 hPa is opposite to that in ERA-40 and the CTL run (Figs. 5b,e). The difference in the horizontal advection will be discussed in sections 4c and 4d. For the vertical advection above 700 hPa (Fig. 5i), the positive anomalies during phases 3–4 and negative anomalies during phases 6–8 are opposite to the results in both ERA-40 and CTL. Below 700 hPa, since the westward tilt is more obvious, the vertical advection peaks at the mature phase and remains positive until phase 7. Therefore the positive anomalies above 700 hPa during phases 3–4 and below 700 hPa after phase 5 directly lead to the slow diminishing of MSE in Fig. 4c; this will be discussed further in Fig. 6.
Height–phase cross section of vertical velocity anomalies on the MJO scale (contours; −1 × 10−2 Pa s−1) and mean MSE (colors; K) averaged over (a) 10°S–10°N, 80°–90°E in ERA-40 and 10°S–10°N, 120°–130°E in the (b) CTL and (c) NSC runs. (d) Vertically integrated total advection from 1000 to 100 hPa (W m−2).
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
As vertical advection is a dominant term in the MSE budget, we further examine it to understand the differences among the ERA-40 data and the CTL and NSC runs. It is found that advection of the mean MSE by vertical velocity perturbations on the MJO scale is the major contributor to the MSE advection (not shown). Figure 6 shows the vertical velocity field during the MJO life cycle, superposed by the mean MSE for ERA-40, CTL, and NSC, respectively. For ERA-40 (Fig. 6a), the midtroposphere during phases 1 and 2 is occupied by downward motion. However, there is upward motion in the lower troposphere during phases 2 and 3, with a maximum below 700 hPa. The mean MSE has a minimum above 700 hPa. Thus, the midtropospheric downward motion leads to positive MSE advection (Fig. 5c). The tilt of upward motion shows that it starts below 900 hPa at phase 1. As the MJO evolves, the center of upward motion moves up from the boundary layer to 400 hPa after phase 4, peaking at phase 5. Accordingly, as the mean MSE increases with height above 700 hPa, upward motion in the midtroposphere leads to negative MSE advection during phases 3–6 (Fig. 5c).
In the CTL run (Fig. 6b), the vertical velocity field shows very similar evolution to the ERA-40, including the feature that upward motion first starts in the boundary layer at early stages of the MJO with its maximum center gradually moving up to 400 hPa after phase 4, reaching maximum at phase 5. Compared to ERA-40 (Fig. 6a), the mean MSE has a broader, lower minimum. The magnitude of vertical velocity is comparable between ERA-40 and CTL. However, the stronger vertical gradient from the lower minimum mean MSE in the CTL leads to larger vertical advection of MSE below 700 hPa than in ERA-40 (Figs. 5c,f).
For the NSC run, there are striking differences from the ERA-40 and CTL results (Fig. 6c). First, the amplitudes of vertical velocity anomalies are much weaker, about half of those in ERA-40 and CTL (Figs. 6a,b). Second, although the maximum upward motion also occurs at phase 5, the transition from downward to upward motion during the developing stage above the PBL occurs much later. The absence of upward motion above the PBL during phases 2 and 3 cannot precondition the lower and middle troposphere for ensuing deep convection. It is apparently a consequence of shutting off shallow convection. Wu (2003) showed in a dynamic model that shallow convection is more efficient in generating low-level convergence and upward motion than deep convection. The downward motion during phases 3–4 above 700 hPa, together with the minimum center of mean MSE near 700 hPa, leads to the positive vertical advection in Fig. 5i. Third, unlike the results of ERA-40 and CTL run that the upward motion decreases rapidly after the mature phase (Figs. 6a,b), there is still large upward motion in the whole troposphere at phases 6 and 7 (Fig. 6c). Also note that in the lower (below 700 hPa) troposphere maximum upward motion occurs at phase 4 in both ERA-40 and CTL, whereas in NSC it occurs at phase 5, producing large positive vertical MSE advection, as seen in Fig. 5i.
Figure 6d shows the evolution of vertically integrated total (vertical and horizontal) advection of MSE from 1000 to 100 hPa, which is the negative of GMS (Raymond and Fuchs 2007). Both the ERA-40 and simulations show that the integrated advection is positive during the beginning and developing stages (phases 1–3) and ending phases (phases 8–9) and negative during strong convection (phases 4–7). Thus, there is net import of MSE into the atmospheric column (negative GMS) before and after the deep convection, which contributes to building up energy for convection. At the mature phase (phase 5), the net export of MSE out of the atmospheric column with maximum GMS releases convective energy. The major difference among the ERA-40 and the simulations is the weaker amplitude of GMS in the NSC run and delayed peaking at phase 6.
c. Moisture budget
In the tropical atmosphere, temperature perturbations are small due to a large Rossby radius, and the MSE changes are dominated by contributions from moisture perturbations (Maloney 2009; Kiranmayi and Maloney, 2011; Andersen and Kuang 2012). Thus, our following analysis examines the moisture evolution and its budget results.
Figure 7 shows the vertical profiles of specific humidity (i.e., q) averaged over the specified areas at different phases, and the evolution is consistent with the changes of MSE anomalies shown in Fig. 4. In the ERA-40 (Fig. 7a), the negative anomalies occur at phases 1, 2, 7, 8, and 9, corresponding to the beginning and ending phases in the MJO life cycle. The positive anomalies are seen at the developing, mature, and early decaying phases (phases 3, 4, 5, and 6). The negative anomalies in the lower troposphere reach a maximum at phase 8 and begin to weaken at phases 9 and 1. By phase 2, positive anomalies have developed in the lower troposphere, with a maximum near 800 hPa. The positive anomalies deepen further at phase 3, with the maximum moving upward. At phase 4 before the mature phase, the positive anomalies peak at 700 hPa. As the MJO evolves into its mature stage (phase 5), the positive moisture anomalies with a maximum near 500 hPa have started to weaken in the lower troposphere, presumably due to moisture depletion by convection. As the MJO decays, dry anomalies appear in the lower troposphere at phase 6.
Vertical profile of specific humidity (g kg−1) on the MJO scale at each phase averaged over (a) 10°S–10°N, 80°–90°E in ERA-40 and 10°S–10°N, 120°–130°E in the (b) CTL and (c) NSC runs.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
The CTL run shows a similar evolution of the vertical structure of moisture anomalies except with weaker amplitude (Fig. 7b), consistent with the weaker MJO signal and MSE anomalies (Figs. 3b and 4b). The dry anomalies are strongest at phase 1 at 500 hPa. There is strong moistening from phase 1 to phase 2. Similar to the ERA-40, positive moisture anomalies develop near 850 hPa at phase 2. At phase 3, positive anomalies continue to expand upward with a peaking moving upward. By phase 4 the maximum has moved up to 700 hPa, as in ERA-40. At the mature phase, the moisture anomalies peak at 500 hPa, corresponding to active deep convection. Meanwhile, dry anomalies start to develop below 900 hPa. By phase 6, large dry anomalies have developed below 700 hPa. The gradual upward expansion of moist anomalies during phases 2–4 in both ERA-40 and CTL clearly helps to precondition the troposphere for deep convection. Similarly, the dry anomalies in the lower troposphere after phase 5 help to shut off the moisture supply for convection.
When shallow convection is turned off (Fig. 7c), there are major differences from the CTL run. Although the moistening starts from phase 2 and then intensifies, the maximum is confined to below 800 hPa and the amplitude is much smaller. The shallow and weak moistening in the lower troposphere before deep convection cannot provide a favorable condition for further development of deep convection in the NSC run. Because of the weak deep convection, the depletion of moisture is also slow after the mature phase. For instance, 500-hPa moisture anomalies at phase 6 are of the same amplitude as those at phase 5, and there are still positive anomalies above 700 hPa at phase 7, while there are large negative anomalies in the lower half of the troposphere in both ERA-40 and CTL.
Comparing with the ERA-40, both CTL and NSC runs have much smaller moisture anomalies in the PBL below 900 hPa. This appears to indicate that the PBL parameterization may be at fault for the lack of response to MJO. Another possibility is the prescribed sea surface temperature for the model simulations. Previous studies have shown significant intraseasonal variability in both the atmosphere and SST, and there might be a positive feedback between them (Li and Wang 1994). As the PBL is coupled to SST through surface turbulent fluxes, the use of monthly mean SST without intraseasonal variation may have contributed to the smaller moisture variability in the PBL on intraseasonal time scales (Waliser et al. 1999; Benedict and Randall 2011).
Next we calculate each term in the moisture budget equation [Eq. (2)] to better understand the contribution of the physical and dynamic processes to the moisture anomalies and try to explain the difference between the two runs. Figure 8 shows the phase–height cross section of the q budget terms averaged over specified areas for the ERA-40 and simulations. For the ERA-40, the tendency term (Fig. 8a) shows that, in the first half of the MJO life cycle (phases 1–4), the atmosphere is moistened and, after the mature phase (phases 5–8), the moisture is depleted, which is consistent with the recharge–discharge mechanism in MJO dynamics. Previous studies have suggested the importance of horizontal advection in moisture budget (Maloney 2009; Kiranmayi and Maloney 2011; Benedict and Randall 2007, 2009; Andersen and Kuang 2012). In Fig. 8b, horizontal advection moistens the troposphere before deep convection (phases 1–4) and peaks at 600 hPa. Dry advection expands upward from 900 hPa after phase 3, reaching the maximum at 700 hPa at phase 6, followed by moistening near the surface at phase 7. These features are in agreement with previous studies (Maloney and Hartmann 1998; Benedict and Randall 2007, 2009). The magnitude of vertical advection (Fig. 8c) is larger than the net tendency and horizontal advection. During the phases of suppressed convection, large-scale subsidence leads to drying. From phases 2 to 5, moisture is transported upward to moisten the lower and middle troposphere, with a peak at phase 5 centered between 400 and 500 hPa corresponding to the upward motion center (Fig. 6a). It is also noted that the vertical structure is tilted toward higher MJO phases, implying a westward tilt.
Height–phase cross section of each term (K day−1) for the q budget (multiplied by L/cp) averaged over (top) 10°S–10°N, 80°–90°E in ERA-40 and 10°S–10°N, 120°–130°E in the (middle) CTL and (bottom) NSC runs: (a),(d),(g) total tendency, (b),(e),(h) horizontal advection, and (c),(f),(i) vertical advection. The regions above 90% confidence level are stippled.
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
In the CTL run, similar to the ERA-40 data, there is net moistening in the first half and drying in the second half of the MJO life cycle (Fig. 8d). For horizontal advection (Fig. 8e), although there is moistening before the mature phase and drying during and after it above 800 hPa with a peak at 600 hPa, the evolution below 800 hPa is opposite, in disagreement with the ERA-40 data. The CTL run simulates well the changes of vertical advection but with a more upright vertical structure, which can be attributed to that of the vertical velocity (Fig. 6b). Similar to the MSE budget, the amplitude of each moisture budget term is larger compared with that in the ERA-40.
For the NSC run, although the tendency term also reflects the recharge–discharge paradigm, the moistening and drying have smaller magnitudes and the maximum center is located at 850 hPa. It is largely attributed to the biases in the horizontal advection, which has a maximum center at 775 hPa (Fig. 8h) compared with that in ERA-40 and the CTL run (Figs. 8b,e), which show maximum horizontal advection at 600 hPa. This will be discussed in section 4d. Since there is still large upward motion in the whole troposphere after mature phase (Fig. 6c), unlike the results of ERA-40 and the CTL run, the vertical advection continues to moisten the troposphere after phase 5 (Fig. 8i), which contributes to the slow decay of positive moisture anomalies (Fig. 7c).
d. Horizontal advection
As noted above, the contribution of horizontal advection to the moisture budget reaches maximum at 600 hPa in ERA-40 and CTL, but at 775 hPa in NSC. (Figs. 8b,e,h). We found that the advection of mean moisture by horizontal wind perturbations on the MJO scale is the major contributor to the horizontal advection (not shown), as also noted in previous studies (Maloney 2009; Kiranmayi and Maloney 2011). Figure 9 shows the 600-hPa horizontal winds on the MJO scale superposed by the mean 
The 600-hPa horizontal winds on the MJO scale (vectors; m s−1) and mean 
Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00127.1
In the ERA-40, the maximum mean moisture is centered at east of 90°E over the equator, the easterly winds bring moisture to the area of interest at phase 2 (Fig. 9a), and the westerly winds lead the dry advection at phase 6 (Fig. 9b). Although the meridional winds are much weaker than the zonal winds, since the meridional moisture gradient is stronger, the divergent winds from the equator contribute to the moistening at phase 2. Conversely, at phase 6 both the meridionally convergent winds and the westerly flow contribute to the dry advection.
In the CTL run, since the maximum MJO convection is in the western Pacific, there are two centers of mean moisture located to the east and west of the averaging area for Fig. 8. At phase 2, the easterly zonal wind and the divergent meridional winds lead to moist advection (Fig. 9c). At phase 6 (Fig. 9d), the moistening from the western maximum center along with westerly winds is confined to 4°S–2°N, whereas in higher latitudes the northerly wind component leads to dry advection.
For the NSC run (Figs. 9e,f), although the structure is similar to that in the CTL (Figs. 9c,d), the gradient of mean moisture and the wind perturbation are much weaker, in particular the zonal component. Therefore the biases in both the mean specific humidity and winds on the MJO scale contribute to the weak horizontal advection, which directly leads to weaker and shallow development of moisture and MSE at the developing phases (Figs. 4c and 7c).
5. Summary
This study investigates the effect of shallow convection on moist static energy and moisture budgets of simulated MJO using the NCAR CAM3.0 driven by observed monthly mean SST. Two experiments were carried out: one is the control run (CTL) that uses a revised Zhang–McFarlane convection scheme for deep convection and the Hack scheme for shallow convection and the other is an experiment run (NSC) in which shallow convection by the Hack scheme is turned off below 700 hPa in the tropical belt within 20° from the equator. The results of November–April average are analyzed, and the major findings are summarized below.
Both simulations, with and without shallow convection in the tropics, produce realistic mean OLR and 850-hPa winds (Fig. 1). Without shallow convection, the atmosphere is much drier in the lower troposphere above the PBL, and more moist within the PBL, resulting in excessive cloud amount in the PBL due to lack of effective upward transport of moisture by shallow convection (Fig. 2).
During the MJO life cycle, the CTL run shows better skill in representing the location, amplitude, and eastward propagation of convection and the coupled structure between winds and convection than the NSC run does (Fig. 3). The recharge–discharge paradigm in the evolution of MSE is reproduced reasonably well in the CTL run except with a smaller amplitude of anomalies compared with that in the ERA-40 (Figs. 4a,b). Without shallow convection, the MJO is substantially weaker, the development of MSE anomalies is confined to much lower levels at the developing phases, and the decay of positive MSE anomalies after the mature phase is slow (Fig. 4c).
The CTL run simulates all MSE budget terms and their evolution well although some quantitative differences exist, including relatively large positive MSE advection in the PBL at the mature phase of MJO in CTL (Fig. 5e). In the NSC run, the biases of MSE evolution can be largely attributed to the weak and shallow development of horizontal advection at the developing phases and the positive biases of vertical advection anomalies above 700 hPa during phase 3–4 and below 700 hPa after the mature phase (Figs. 5h,i). The vertical advection differences between the CTL and NSC is directly due to differences in MJO-scale vertical velocity anomalies. While CTL reproduces the gradual transition from downward to upward motion during developing phases (phases 2–4) of the MJO (Fig. 6b), the vertical velocity in NSC is dominated by downward motion in phases 2 and 3 and then abruptly transitions to upward motion at phase 4, indicating a lack of preconditioning (Fig. 6c). The gradual transition in CTL and abrupt transition in NSC of vertical velocity simply reflect the fact that shallow convection is more efficient in generating low-level convergence and upward motion than deep convection (Wu 2003).
The moisture field shows a major difference with or without shallow convection. Both ERA-40 data and the CTL simulation show a gradual, extensive moistening in the lower troposphere before deep convection (Figs. 7a,b), in good agreement with many previous observations (Maloney and Hartmann 1998; Tian et al. 2006; Benedict and Randall 2007). Without shallow convection, besides much smaller amplitudes of moisture anomalies, most of the moistening is confined to below 800 hPa in the developing phases, indicating that the insufficient moistening in the lower troposphere during this stage prevents the development of deep convection (Fig. 7c).
Similar to the MSE budget, the major biases of the moisture budget in the NSC run is largely due to vertical advection before and after the mature phase and horizontal advection at the developing phases. The downward motion in the developing phases prevents moistening above the PBL, and the lingering upward motion after the mature phase slows the drying (Figs. 6c and 8i). For horizontal advection, although the NSC run shows moist advection in the developing stage and dry advection in the decay stage of MJO, as in the observations (Benedict and Randall 2007) and ERA-40, the maximum centers are at much lower levels. The weaker gradient of mean moisture and the wind perturbation, in particular the zonal component, directly lead to weak horizontal advection at 600 hPa before deep convection (Figs. 8h and 9e).
Acknowledgments
This work was supported by the National Program on Key Basic Research Project (2010CB951904), the National Science Fund for Distinguished Young Scholars (41125017), the China Meteorological Administration (GYHY200706010), and the National Key Technologies R&D Program (2007BAC29B03). GJZ was supported by the National Science Foundation Grant AGS-1015964 and National Oceanic and Atmospheric Administration Grants NA08OAR4320894 and NA11OAR4310098. The computational support was provided by the NCAR CISL facilities.
REFERENCES
Andersen, J. A., and Z. M. Kuang, 2012: Moist static energy budget of MJO-like disturbances in the atmosphere of a zonally symmetric aquaplanet. J. Climate, 25, 2782–2804.
Benedict, J. J., and D. A. Randall, 2007: Observed characteristics of the MJO relative to maximum rainfall. J. Atmos. Sci., 64, 2332–2354.
Benedict, J. J., and D. A. Randall, 2009: Structure of the Madden–Julian oscillation in the superparameterized CAM. J. Atmos. Sci., 66, 3277–3296.
Benedict, J. J., and D. A. Randall, 2011: Impacts of idealized air–sea coupling on Madden–Julian oscillation structure in the superparameterized CAM. J. Atmos. Sci., 68, 1990–2008.
Bladé, I., and D. L. Hartmann, 1993: Tropical intraseasonal oscillations in a simple nonlinear model. J. Atmos. Sci., 50, 2922–2939.
Bretherton, C. S., M. E. Petters, and L. E. Back, 2004: Relationships between water vapor path and precipitation over the tropical ocean. J. Climate, 17, 1517–1528.
Collins, W. D., and Coauthors, 2006: The formulation and simulation of the Community Atmosphere Model: CAM3. J. Climate, 19, 2144–2161.
Del Genio, A. D., Y. Chen, D. Kim, and M. S. Yao, 2011: The MJO transition from shallow convection to deep convection in CloudSat/CALIPSO data and GISS GCM simulation. J. Climate, 25, 3755–3770.
Duchon, C. E., 1979: Lanczos filtering in one and two dimensions. J. Appl. Meteor., 18, 1016–1022.
Fuchs, Z., and D. J. Raymond, 2005: Large-scale modes in a rotating atmosphere with radiative–convective instability and WISHE. J. Atmos. Sci., 62, 4084–4094.
Grabowski, W. W., and M. W. Moncrieff, 2004: Moisture–convection feedback in the tropics. Quart. J. Roy. Meteor. Soc., 130, 3081–3104.
Hack, J. J., 1994: Parameterization of moist convection in the National Center for Atmospheric Research Community Climate Model (CCM2). J. Geophys. Res., 99, 5551–5568.
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.
Inness, P. M., J. M. Slingo, S. J. Woolnough, R. B. Neale, and V. D. Pope, 2001: Organization of tropical convection in a GCM with varying vertical resolution: Implications for the simulation of the Madden–Julian oscillation. Climate Dyn., 17, 777–793.
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.
Kemball-Cook, S. R., and B. C. Weare, 2001: The onset of convection in the Madden–Julian oscillation. J. Climate, 14, 780–793.
Kim, D., and Coauthors, 2009: Application of MJO simulation diagnostics to climate models. J. Climate, 22, 6413–6436.
Kiranmayi, L., and E. D. Maloney, 2011: Intraseasonal moist static energy budget in reanalysis data. J. Geophys. Res., 116, D21117, doi:10.1029/2011JD016031.
Li, C., X. Jia, J. Ling, W. Zhou, and C. Zhang, 2009: Sensitivity of MJO simulations to convective heating profiles. Climate Dyn., 32, 167–187, doi:10.1007/s00382-008-0455-x.
Li, T. M., and B. Wang, 1994: The influence of sea surface temperature on the tropical intraseasonal oscillation: A numerical study. Mon. Wea. Rev., 122, 2349–2362.
Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 1275–1277.
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.
Maloney, E. D., 2009: The moist static energy budget of a composite tropical intraseasonal oscillation in a climate model. J. Climate, 22, 711–729.
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.
Maloney, E. D., and D. L. Hartmann, 2001: The sensitivity of intraseasonal variability in the NCAR CCM3 to changes in convective parameterization. J. Climate, 14, 2015–2034.
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:10.1029/2005JD007003.
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:10.1029/2007JD008700.
Peters, O., and J. D. Neelin, 2006: Critical phenomena in atmospheric precipitation. Nat. Phys., 2, 393–396, doi:10.1038/nphys314.
Rasch, P. J., and J. E. Kristjánsson, 1998: A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations. J. Climate, 11, 1587–1614.
Raymond, D. J., and Z. Fuchs, 2007: Convectively coupled gravity and moisture modes in a simple atmospheric model. Tellus, 59A, 627–640.
Raymond, D. J., and Z. Fuchs, 2009: Moisture modes and the Madden–Julian oscillation. J. Climate, 22, 3031–3046.
Raymond, D. J., S. L. Sessions, A. H. Sobel, and Z. Fuchs, 2009: The mechanics of gross moist stability. J. Adv. Model. Earth Syst., 1 (9), doi:10.3894/JAMES.2009.1.9.
Salby, M. L., and H. H. Hendon, 1994: Intraseasonal behavior of clouds, temperature, and motion in the tropics. J. Atmos. Sci., 51, 2207–2224.
Simmons, A. J., and J. K. Gibson, 2000: The ERA-40 project plan. ERA-40 Project Report Series, Vol. 1, ECMWF, 63 pp.
Stevens, B., 2007: On the growth of layers of nonprecipitating cumulus convection. J. Atmos. Sci., 64, 2916–2931.
Tian, B., D. E. Waliser, E. J. Fetzer, B. H. Lambrigtsen, Y. L. Yung, and B. Wang, 2006: Vertical moist thermodynamic structure and spatial–temporal evolution of the MJO in AIRS observations. J. Atmos. Sci., 63, 2462–2485.
Tian, B., D. E. Waliser, E. J. Fetzer, and Y. L. Yung, 2010: Vertical moist thermodynamic structure of the Madden–Julian oscillation in atmospheric infrared sounder retrievals: An update and a comparison to ECMWF Interim Re-Analysis. Mon. Wea. Rev., 138, 4576–4582.
Waliser, D. E., K. M. Lau, and J.-H. Kim, 1999: The influence of coupled sea surface temperatures on the Madden–Julian oscillation: A model perturbation experiment. J. Atmos. Sci., 56, 333–358.
Wang, B., 1988: Dynamics of tropical low-frequency waves: An analysis of the moist Kelvin wave. J. Atmos. Sci., 45, 2051–2065.
Wheeler, M., 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.
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.
Zhang, C., 2005: Madden–Julian oscillation. Rev. Geophys., 43, RG2003, doi:10.1029/2004RG000158.
Zhang, G. J., 2002: Convective quasi-equilibrium in midlatitude continental environment and its effect on convective parameterization. J. Geophys. Res., 107, 4220, doi:10.1029/2001JD001005.
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.
Zhang, G. J., and M. Mu, 2005a: Simulation of the Madden–Julian oscillation in the NCAR CCM3 using a revised Zhang–McFarlane convection parameterization scheme. J. Climate, 18, 4046–4064.
Zhang, G. J., and M. Mu, 2005b: 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:10.1029/2004JD005617.
Zhang, G. J., and X. L. Song, 2009: Interaction of deep and shallow convection is key to Madden–Julian oscillation simulation. Geophys. Res. Lett., 36, L09708, doi:10.1029/2009GL037340.
Zhang, M., W. Lin, C. B. Bretherton, J. J. Hack, and P. J. Rasch, 2003: A modified formulation of fractional stratiform condensation rate in the NCAR Community Atmosphere Model (CAM2). J. Geophys. Res., 108, 4035, doi:10.1029/2002JD002523.
