1. Introduction
The Madden–Julian oscillation (MJO) is the dominant mode of subseasonal variability during the boreal winter–spring (Madden and Julian 1971, 1972, 1994). During years when it is active, the MJO tends to make several apparent circuits of the globe. Over the tropical Eastern Hemisphere the MJO controls convection on time scales of 30–70 days, and it is typically associated with reduced seasonal monsoon rainfall over northern/central Australia (Hendon et al. 1999). Because of the large-scale nature and temporal persistence of the convection, extratropical teleconnections have been noted (e.g., Weickmann et al. 1985; Murakami 1988), including an influence on rainfall over the western United States (Mo and Higgins 1998) and an enhancement of skill in the extratropics in medium- and extended-range numerical weather prediction (Ferranti et al. 1990). In the western Pacific the MJO convection tends to propagate into the South Pacific convergence zone, and in the central Pacific it ceases. Even so, the MJO signal continues to be manifested in the global circulation as a dry Kelvin wave, having been seen in the tropical 200-hPa velocity potential and zonal wind (e.g., Slingo et al. 1996), and more recently as a sea level pressure surge that impacts the Andes and the East African highlands (Matthews 2000).
Examination of near-surface variables has established the importance of the exchange of heat, moisture, and momentum at the air–sea interface during the eastward propagation of the MJO (Zhang and McPhaden 1995; Lin and Johnson 1996; Zhang 1996; Flatau et al. 1997; Lau and Sui 1997; Sperber et al. 1997; Shinoda et al. 1998, 1999; Wang and Xie 1998; Woolnough et al. 2000). Shinoda et al. (1999) have validated surface flux anomalies from National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis against point measurements from the Improved Meteorological Instrument (IMET; Hosom et al. 1995) buoy, located at 1.75°S, 156°E. During the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE; 24 October 1992–28 February 1993), three strong MJOs passed over the IMET buoy. They found the phase and amplitude of the reanalyzed net surface heat flux, the latent heat flux, and zonal wind stress to agree “reasonably well” with the IMET measurements. However, the surface shortwave and longwave variations were less well captured, though their phasing was realistic in MJO events during 1986–93.
Numerous mechanisms for the eastward propagation of the MJO have been suggested. As originally proposed, the evaporation wind feedback (also known as wind-induced surface heat exchange) of Emanuel (1987) and Neelin et al. (1987) is not the operative mechanism that promotes the eastward propagation of the MJO. This arises because of the mismatch in the location of the strongest surface fluxes. Observations suggest wave conditional instability of the second kind (CISK; Lau and Peng 1987; Wang and Xie 1988), augmented by surface friction (Hendon and Salby 1994), is the dominant mechanism for the eastward propagation of the MJO. In this scenario, equatorial low-level convergence leads the convection (e.g., Rui and Wang 1990) and is in phase with the buildup of moisture in the boundary layer (e.g., Hendon and Salby 1994; Jones and Weare 1996; Maloney and Hartmann 1998). This scenario applies to periods when the MJO convection is established over the central Indian Ocean and the west/central Pacific Ocean. Alternatively, Kemball-Cook and Weare (2001) examined off-equatorial station-sonde data, finding that while moistening does occur in advance of the convection, it is not due to low-level convergence. Their results are consistent with the “discharge–recharge” mechanism of Bladé and Hartmann (1993) in which the time scale of the MJO is determined by local thermodynamic processes that govern the development of convective instability.
The baroclinic nature of the MJO has been elucidated (e.g., Madden and Julian 1971; Knutson and Weickmann 1987). However, studies have typically been limited to examination of one upper-tropospheric and one lower-tropospheric level, and the interrelationships as a function of altitude have not been explored in detail. Furthermore, little attention has been devoted to the conditions occurring during the onset of MJO convection in the western Indian Ocean. The purpose of this paper is to examine in detail the vertical structure of the MJO and to provide a comprehensive picture of the MJO within the dynamically consistent framework of the NCEP–NCAR reanalysis. This in turn will provide a more comprehensive suite of diagnostics for understanding the limited ability of general circulation models to simulate the MJO (e.g., Slingo et al. 1996; Sperber et al. 1997).
The data used in this study are described in section 2; in section 3 the MJO is described, including the sensitivity of the dominant modes of convection to the years analyzed, the propagation, vertical structure, and onset of convection; and in section 4 discussion and conclusions are given.
2. The data
The NCEP–NCAR reanalysis was a joint project between NCEP and NCAR to produce a multidecadal record of global atmospheric analyses with a fixed data assimilation system (Kalnay et al. 1996). The data assimilation and forecast model was implemented operationally at NCEP in January 1995. The model is run at a horizontal resolution of T62, with 28 vertical levels. Moist convection is represented by a simplified Arakawa–Schubert parameterization scheme (Pan and Wu 1994), and clouds are diagnosed using a scheme based on Slingo (1987). The NCEP model uses the three-layer soil scheme of Pan and Mahrt (1987) in which the temperature of the bottom layer is set to the annual mean climatological value. Data were assimilated using a spectral statistical interpolation–3D variational analysis method that requires no nonlinear normal mode initialization. Upper-air data on standard pressure surfaces have been supplied on a 2.5° latitude–longitude grid. Surface and 24-h forecast fields (e.g., fluxes) are on the equivalent T62 Gaussian grid. The spinup of the hydrological cycle is small in the NCEP–NCAR reanalysis (e.g., Mo and Higgins 1996; Stendel and Arpe 1997). For the period analyzed herein, the optimally interpolated weekly sea surface temperatures (SSTs) of Reynolds and Smith (1994) were linearly interpolated to daily values.
Subseasonal variations of rainfall are characterized using the Climate Prediction Center Merged Analysis of Precipitation (CMAP). This dataset uses essentially the same algorithm and data sources as the monthly CMAP dataset described by Xie and Arkin (1997). The version used is based on a blend of gauge data with satellite products, including Geostationary Operational Environmental Satellite (GOES) Precipitation Index (GPI; based on geostationary infrared data), Microwave Sounding Unit (MSU), outgoing longwave radiation–Based Precipitation Index (OPI), Special Sensor Microwave Imager (SSMI) scattering, and SSMI emission.
The Advanced Very High Resolution Radiometer outgoing longwave radiation (AVHRR OLR) on the National Oceanic and Atmospheric Administration (NOAA) polar orbiting satellites has been used to identify the convective signature of the MJO. These data have been daily averaged and processed on a 2.5° latitude–longitude grid, with missing values filled by interpolation (Liebmann and Smith 1996). This dataset has been used in a variety of MJO studies (e.g., Salby and Hendon 1994; Slingo et al. 1999) and is a reasonable proxy for tropical convection (Arkin and Ardanuy 1989).
3. The Madden–Julian oscillation
a. MJO index and convective structure
The large-scale convective signal of the MJO gives rise to strong perturbations in the zonal wind and thus affects the length of day through its influence on the earth's angular momentum budget (Rosen and Salstein 1983; Madden 1988). Using the zonal mean zonal wind, averaged between 10°N and 10°S, Slingo et al. (1996, 1999) devised an index of intraseasonal variability. This time series is bandpass filtered using a 20–100-day Lanczos filter, and the envelope of intraseasonal variability is isolated by calculating the variance in a 101-day running window (Fig. 1). This index captures the preferred seasonality of the MJO, being most active during the boreal winter–spring. It also highlights the irregular nature of the MJO on interannual time scales, which is not controlled by the phase of ENSO (Slingo et al. 1999). Among alternate MJO indices (including those based on wavenumber-1 and wavenumber-2 diagnostics) this index “appears to delineate the periods of high MJO activity more successfully than the other more regionally PC [principal component]-based indices” (Slingo et al. 1999).
Using the index in Fig. 1, strong MJO activity is isolated. Heretofore, single or successive years of data have been analyzed typically (e.g., Sperber et al. 1997), though in some instances techniques for isolating individual events have been employed (e.g., Shinoda et al. 1998; Woolnough et al. 2000). As seen in Fig. 1, using consecutive years of data may result in the inclusion of years with little or no MJO variability. While this latter approach still captures many of the features of the MJO and its interactions, concentrating on years of strong MJO activity gives a more accurate portrayal of its behavior, as will be seen in the forthcoming analysis.
The lead–lag relationship and the dominant convective signature of the MJO is investigated using empirical orthogonal function (EOF) analysis on 20–100-day bandpass-filtered AVHRR OLR, concentrating on the months November–March, when the MJO tends to be strongest. As discussed in Salby et al. (1994), this is the time of year with the greatest amplification of the Kelvin wave and the forced Rossby gyres, occurring when the SST and atmospheric heating tend to be at a maximum near the equator. EOFs 1 and 2 from the analysis using nine successive winters/springs of data (1982–83 through 1990–91) are given in Figs. 2a and 2b. Comparatively, in Figs. 2f and 2g the EOFs from seven years of strong MJO activity (Fig. 1; 1984–85, 1985–86, 1987–88, 1989–90, 1991–92, 1994–95, and 1996–97)1 each account for a larger fraction of the explained variance. As seen in Fig. 2f, during years of strong MJO activity the enhanced convection near the Maritime Continent is stronger, and the longitudinal extent is larger than that seen in Fig. 2a for the consecutive-year analysis. For EOF-2 there is a stronger projection of MJO variability near the date line during strong years (Fig. 2g versus Fig. 2b), and the local percent variance explained by the first two EOFs is greater (Fig. 2h versus Fig. 2c). For both choices of data period, the maximum positive correlation indicates that PC-2 leads PC-1 by 12 days on average, but during strong MJO years the maximum correlation is 0.83 rather than 0.60 for the consecutive years (Figs. 2i and 2d). This indicates a greater degree of coherence in the phasing of PC-1 and PC-2 and thus the eastward propagation of convection during the strong years. This is consistent with the closer agreement of their power spectra and the greater power in the intraseasonal band relative to consecutive years (Figs. 2j and 2e).
b. The space–time evolution of the MJO
Numerous authors have described aspects of the space–time structure of the MJO (e.g., Weickmann et al. 1985; Hendon and Salby 1994; Woolnough et al. 2000), but detailed analysis of the vertical structure has been lacking. Furthermore, previous analyses have concentrated on analyzing anomalies, but here the total field is also considered, since the background state upon which the anomalies project is of importance.
Observations suggest that frictional wave-CISK (low-level moisture convergence) is responsible for the eastward propagation of the MJO (e.g., Hendon and Salby 1994; Jones and Weare 1996). However, the relationship of the convection to the surface fluxes changes during the evolution of the MJO, as illustrated in Fig. 3, based on the “models” of Zhang and McPhaden (2000). Model I tends to be present in the Indian Ocean, whereas model II prevails over the western Pacific, with the strongest fluxes coincident with maximum low-level westerlies and the convection.
To explore the space–time structure of the MJO, the PC-1 time series from Fig. 2i is linearly regressed against numerous variables. The regressions have been calculated for lags of ±25 days. For lag 0, 150 days of data are used, spanning 1 November–30 March for each year (29 March during leap year). For the lag regressions, additional target data is extracted so that the regressions include 150-day periods for each year. For example, in the case of the PC-1 leading by 5 days, the target field analyzed would span the period 6 November–4 April (3 April during leap year). The regressed fields are then scaled by a one-standard-deviation perturbation of the PC and plotted where the regression exhibits at least 5% significance. The 5% significance level is calculated assuming each pentad is independent. This is a reasonable choice since 1) the size of the convective envelope is consistent with that observed in satellite images during the active (convective) phase of the MJO and 2) it is more conservative than calculating the degrees of freedom using the full autocorrelation structure of the data (Livezey and Chen 1983), which is computationally much more expensive.
Jones et al. (1998) presented a lag correlation analysis for the period January 1985–April 1991, estimating the shortwave radiation using the technique of Gautier et al. (1980) based on input from the International Satellite Cloud Climatology Project (Rossow et el. 1988), and used bulk formula to estimate the latent heat flux using European Centre for Medium-Range Weather Forecasts (ECMWF) analyses. This provides a point of comparison with which to check consistency between their and my results. However, a benefit of the present analysis is the assessment of the interactions occurring in the western–central Pacific Ocean, which were not readily captured in the Jones et al. (1998) analysis since their reference time series, based in the central Indian Ocean, did not have a strong projection onto convection there. Woolnough et al. (2000) is an additional comparison point based on the 15-yr ECMWF reanalysis (ERA-15).
1) Lag 0 spatial and vertical structure
Linear regression at zero time lag is used to show the large-scale structure of the MJO. Figures 4 and 5 show the regressions of PC-1 against the 20–100-day bandpass-filtered data and the total data, respectively. Figures 4a and 4b indicate the close correspondence between the reduced AVHRR OLR and the enhanced CMAP rainfall in the vicinity of the Maritime Continent. Consistent with EOF-1 (Fig. 2f), the OLR is reduced in excess of 10 W m−2 over this region in association with enhanced rainfall of 2–5 mm day−1. Regressions against the total OLR and rainfall (Figs. 5a and 5b) indicate that the deep convection/strongest rainfall is confined to the region 5°N–12.5°S, 95°–165°E. Importantly, the 500-hPa vertical velocity regressions from the reanalysis (Figs. 4c and 5c) are consistent with the AVHRR OLR and the CMAP rainfall estimates. Also, the reduced net shortwave radiation at the surface is collocated with the enhanced convection (Fig. 4d), with the lowest net shortwave occurring just north of the equator (Fig. 5d).
The baroclinic wind structure is seen in Figs. 4a,b and 5a,b. The 200-hPa outflow is dominated by easterlies that extend to the west from the leading edge of the convection. At 850 hPa westerlies dominate at and west of the convection. In the total field, the upper-level (low-level) easterlies (westerlies) are more closely confined to the equatorial region. The surface winds and wind stress closely mimic the behavior at 850 hPa (Figs. 4e,f and 5e,f).
Over the Indian Ocean and Maritime Continent the westerlies enhance the climatological circulation (not shown), giving rise to enhanced latent heat flux (evaporative cooling; Fig. 4f). From 150° to 180°E the near-equatorial easterlies weaken the climatological westerly flow, resulting in a weaker than normal latent heat flux. As seen in Fig. 4g, the pattern of the net surface heat flux is similar to the latent heat flux, with the evaporative cooling accounting for upward of 50% of the net surface heat flux anomalies (Fig. 4h). The lowest contribution is over the Maritime Continent, where the net surface shortwave radiation contributes substantially to the reduced net surface heat flux. There, the contribution of the latent heat flux to the net surface heat flux is likely an overestimate, since Shinoda et al. (1999) found that the reanalysis underestimates the shortwave contribution by a factor of 2. In the total, Fig. 5f indicates the latent heat flux to be strongest just north of Australia and in the Indian Ocean. As seen in Fig. 4e, this is associated with a reduction in SST of ∼0.15 K to the west of the convection, while to the east of the convection the SST is ∼0.1 K above normal. The negative (positive) SST signal is displaced to the west (east) of the convective center (125°E), where the westerly (easterly) anomalies were established at earlier time lags. The amplitude of the SST variation is an underestimate since the reanalysis SST was based on pentad data. Also, the regression yields more modest values given the large number of events analyzed. For example, during TOGA COARE2 the amplitude of intraseasonal SST fluctuations associated with individual MJO events was about 1.0 K (Lau and Sui 1997). As seen in Fig. 5e, over both regions the SST is above the 28°C threshold typically associated with deep convection, though that in the western Pacific is warmest, being in excess of 29.25°C. That convection does not extend farther west and east over the warm water is due to the zonal circulation environment within which it is situated [see section 3b(3)]. From Figs. 6a and 6b we note that the enhanced moisture and convergence (negative divergence) near the equator extends farther east than the enhanced rainfall, consistent with the low-level moisture convergence paradigm.
Composite spatial maps of a subset of these fields have been presented by Woolnough et al. (2000). They composited based on OLR at 82.5° and 162.5°, corresponding to approximately day −15 and day 10 relative to PC-1, respectively (not shown). The main discrepancy between the results presented herein and the ERA-15 results is that the NCEP–NCAR reanalysis shortwave signature is weaker over the Indian Ocean compared to the western Pacific. Also, their maximum convective signal occurred at the equator, and different numbers of events were analyzed over the Indian Ocean relative to the western Pacific.
Vertical cross sections at zero time lag are instrumental in depicting the mechanics of the MJO (Fig. 7). At 200 hPa, divergence dominates, as does easterly outflow from the convection. Near the center of the convection (the vertical dashed lines in Figs. 7a–d) the strongest anomalies of convergence and upward velocity occur from 500–300 hPa (Figs. 7a–c). The largest specific humidity enhancement occurs from 700–600 hPa (Fig. 7d) in conjunction with a broad region of enhanced convergence. Here the strongest lateral inflow occurs, with the westerly and easterly wind anomalies each reaching a maximum of approximately |2| m s−1 (Fig. 7c). The midtropospheric easterly anomalies and the enhanced upward motion and moisture to the east of the convective center favor the development of convection, while the westerlies, subsidence, and below-normal moisture to the west of the convective center erode the convective envelope. This indicates a strong role for free-tropospheric processes in modulating the MJO life cycle.
Preconditioning of the lower troposphere east of the center of convection is important. There, enhanced convergence, upward motion, and increased moisture are dominated by signatures that occur increasingly closer to the surface in a given atmospheric column, resulting in a westward tilt with height. These results confirm the modeling study of Woolnough et al. (2001) in which the organization of tropical convection by intraseasonal SST anomalies was investigated using an aqua-planet general circulation model.
The meridional extent of the anomalies at 125°E is seen in Figs. 7e and 7f. Convergence dominates the atmospheric column, particularly near the equator (Fig. 7e). However, it is strongest near the surface at ∼12°S, where the anomalous and total rainfall are strongest (Figs. 4b and 5b) in conjunction with strong anomalous upward vertical velocities from 700–300 hPa (Figs. 7f and 7g). Midtropospheric moistening is also apparent (Fig. 7h), while the lower troposphere is strongly moistened farther south. Even so, near the surface at 20°S (the Great Sandy Desert) the total moisture is only about 0.01–0.0125 kg kg−1, while closer to the equator it is 0.0175–0.02 kg kg−1 (not shown). At the equator, 200-hPa divergence is strongest, extending 20° poleward. The meridional outflow is dominated by southerly anomalies that exhibit a northward tilt, as the Hadley circulation is enhanced (Fig. 7g).
2) MJO propagation
PC-1 regressions against filtered CMAP precipitation and 850-hPa wind, and net surface heat flux and the zonal wind stress for day −20 through day 20, are given in Fig. 8. Those against surface temperature and winds are given in Fig. 9. At day −20 the inactive (dry) phase of the MJO dominates the western Pacific (Fig. 8a). There, increased net surface heat flux (Fig. 8f) contributes to warming the local SST (Fig. 9a). The last vestige of enhanced rainfall from the previous active (convective) phase of the MJO is located near 150°W. Over the Indian Ocean the next active phase develops in the presence of low-level easterly wind anomalies associated with the subsidence farther east. This indicates that the mechanism for the onset/development of MJO convection over the Indian Ocean is different than that associated with its eastward propagation.
By day −15, the Indian Ocean easterlies give way to westerly anomalies and westerly mean winds in conjunction with the intensification of the convection and the cooling in the central Indian Ocean (not shown). The day −10 rainfall over the Indian Ocean enhances further (Fig. 8b), but it exhibits little eastward propagation relative to day −15. Rather, the islands to the east act as an impediment, most likely through their ability to discharge convective available potential energy (Nitta et al. 1992; Wang and Li 1994) because of the strong diurnal land–sea breeze and orographic rainfall. The total (not shown) and anomalous westerly winds now underride the enhanced rainfall. The westerly wind stress anomalies near the trailing edge of the enhanced rainfall increases the evaporative cooling (not shown), resulting in a reduced net surface heat flux (Fig. 8g) and below-normal SST (Fig. 9b). Because of the time delay for establishing the westerlies, the SST anomalies tend to be in quadruture with the center of the convective anomalies. The wind response at the surface (Fig. 9b) is consistent with that at 850 hPa. Farther to the east, the below-normal rainfall affects the South Pacific convergence zone (SPCZ), while in the western equatorial Indian Ocean reduced rainfall is found. A convectively forced Rossby wave is apparent in the eddy stream function (not shown), consistent with Knutson and Weickmann (1987).
From day −5 through day 5 the enhanced rainfall extends eastward to the north of Australia and into the SPCZ. As seen in Fig. 8c for day 0, the fetch of the westerly wind stress has increased the latent heat flux (Fig. 4f), resulting in reduced net surface heat flux over the central Indian Ocean and Maritime Continent (Fig. 8h). By day 10, the convection has moved into the central Pacific and SPCZ (Fig. 8d). The westerly surface stress anomalies extend west into the eastern Indian Ocean and continue to be associated with reduced net surface heat flux anomalies (Fig. 8i) and below-normal surface temperature (Fig. 9d). Over the Indian Ocean the inactive phase of the MJO begins to dominate. It blossoms further by day 20 (Fig. 8e), with net surface heating (Fig. 8j) warming the Indian Ocean (Fig. 9e).
The initial stage of the next active phase of the MJO is found in the equatorial western Indian Ocean, where enhanced rainfall is seen on day 10 (Fig. 8d). This is discussed in more detail in section 3b(4).
3) Longitude–time lag and time lag–pressure plots
The propagation of the MJO can be described further using 5°N–5°S-averaged longitude–lag plots (Fig. 10). For all variables eastward propagation dominates, particularly over the Eastern Hemisphere. Figure 10a shows the reduced AVHRR OLR and enhanced CMAP rainfall anomalies to be closely in phase, and near the Maritime Continent the enhanced convection tends to linger from about day −7 through 7 (note the change in slope of the negative OLR contours). The upward motion at 500 hPa and the net surface shortwave radiation (Fig. 10b and 10c, respectively) are consistent with the observed rainfall and OLR, particularly over the Eastern Hemisphere. The net surface shortwave anomalies are weaker over the Indian Ocean relative to Indonesia and the western Pacific. To the east of the date line, from approximately day −7 through 10, increased upward motion is associated with reduced rainfall and increased OLR, suggesting that the reanalysis may be problematic east of the date line on intraseasonal time scales (see below). Nonetheless, over the Eastern Hemisphere, where the intraseasonal convective signature is strongest, the reanalysis appears to provide a consistent radiative–dynamical framework with which to analyze intraseasonal variations.
As seen in Fig. 10d, the zonal wind stress lags the convection and consequently the enhanced latent heat flux (evaporative cooling) shows a more stunted onset over the Indian Ocean (Fig. 10e). Over the Maritime Continent the latent heat flux comes more closely into phase with the convection since the convection lingers over this region before propagating into the western Pacific. This signifies the transition from model I to model II conditions (Fig. 3). The integrated effect of the increased latent heat flux is a cooling of the SST (Fig. 10f). Figures 10g and 10h further demonstrate the enhanced convergence and moisture prior to convection, consistent with the day 0 results in Figs. 6 and 7 and the low-level moisture convergence paradigm.
Figures 8a–e indicate that the strongest convection occurs south of the equator. Longitude–lag plots for data averaged between 5°–15°S reveal that outside of the near-equatorial region the 1000-hPa divergence anomalies are in phase with the convection anomalies, even though a buildup of low-level moisture precedes the convection (not shown). This is consistent with the radiosonde data analyzed by Kemball-Cook and Weare (2001) and indicates that the convective evolution of the MJO involves other interactions in addition to low-level moisture convergence.
In Fig. 11 the vertical structure of divergence, vertical velocity, winds, and specific humidity at 5°N–5°S, 125°E, as a function of time lag show the development of atmospheric conditions during the life cycle of the MJO with respect to the convective maxima of EOF-1. From approximately day −25 through −15 anomalies of divergence, downward motion, easterly winds, and low specific humidity prevail from the lower troposphere to 300 hPa during the suppressed convective phase of the MJO. However, near the surface, convergence anomalies and upward motion are apparent. As the suppressed convection weakens, moistening of the boundary layer begins at day −16. Subsequently, the convergence and upward motion penetrate higher in the atmosphere, and by about day −8 convection becomes the norm at this location, with moistening of the whole atmospheric column being manifested. That moistening in the lower troposphere leads that in the upper troposphere is consistent with the analysis of Television Infrared Operation Satellite (TIROS) Operational Vertical Sounder (TOVS) specific humidity by Myers and Waliser (2003). This indicates that a realistic moisture signal in the reanalysis is present, at least in terms of phasing relative to the convection, and is consistent with the buildup of moist static energy, which is dominated by the specific humidity (Kemball-Cook and Weare 2001). At this time divergent flow and easterly anomalies characterize the behavior at 200 hPa.
Beginning first at the surface, the easterly anomalies weaken until westerlies dominate to 400 hPa by about day −2 (Fig. 11c). Near day 0 the strongest moistening occurs between 700 and 600 hPa, with the strongest upward motion and divergent outflow being located nearer the tropopause. Near the surface, downward vertical velocity leads the divergence anomalies in the wake of the convection. These anomalies increase in depth and weaken the convection as the MJO enters the suppressed phase.
The results in Fig. 11 show the precursor signals at the surface as a function of time at a fixed location and are consistent with the longitude–height characteristics obtained at day 0 (Fig. 7). This indicates that the potentially problematic inconsistency between vertical velocity and OLR east of the date line noted earlier has not compromised the basic characteristics of the vertical structure of the MJO.
4) Kelvin wave–Rossby wave interactions and the onset of convection in the Indian Ocean
The longitude–lag relationships and vertical structure of the divergence and the specific humidity raise questions regarding the onset mechanism for eastward propagation of the MJO since they appear to circumnavigate the globe (Figs. 10g and 10h). During the convective phase of the MJO, lower pressure lies to the east of higher pressure from the Indian Ocean to the date line (Fig. 12a). As seen in Figs. 10d, 10g, and 10h, near the date line the Kelvin wave decouples from the weakening convection, propagating to the South American coast in about 3 days, with the convergence signal leading the moisture increase. Figure 12a indicates that these signals are associated with the low pressure surge that impacts, and is delayed by, first the Andes and later the East African highlands [consistent with Fig. 6 of Matthews (2000)]. Figure 12b shows that the convergence leads the sea level pressure (SLP) signal over the convectively active region, but east of the date line the convergence and SLP signal are nearly in phase. During the active phase, Fig. 12c indicates that low SLP anomalies lead the westerly anomalies, and for a given time lag the pressures are relatively lower near the leading edge of the westerlies. Adjacent to the west coast of South America, the zonal wind stress anomalies become westerly on about day −3 as the low pressure surge impacts the Andes (Fig. 12c). These wind stress anomalies subsequently exhibit westward propagation as the low pressure surge accumulates near the coast. At about day 10 near 100°W this merges with the eastward propagating westerly anomalies that arose because of the intensified pressure gradient change east of the date line. A similar signal is also noticed for the low-level specific humidity (Fig. 10g). At this time, the low pressure surge becomes established over continental Africa, with high pressure anomalies over the Indian Ocean. This is consistent with the Rossby wave–induced equatorial low-level easterly anomalies and low-level anticyclones to the west of the suppressed convection (Fig. 8d).
Importantly, the aforementioned day 10 conditions are accompanied by enhanced rainfall near 2°N, 55°E (Fig. 8d), as the initial stage of the next active phase of the MJO develops. This signal in the CMAP rainfall is not spurious, as upward vertical velocity anomalies (Fig. 13h) and reduced net surface shortwave anomalies (not shown) are collocated with the rainfall enhancement. The enhanced rainfall occurs in phase with convergence from the surface to 850 hPa (Figs. 13a–d) and enhanced moisture at 925–850 hPa (Figs. 13f and 13g). At this time, conditions over Africa are drier than normal as the eastward propagation of enhanced specific humidity has yet to reach the Eastern Hemisphere (Figs. 10g and 13f–h). As seen in Fig. 13h, the drying extends to 600 hPa near the East African highlands, where subsidence is apparent. This subsidence is mainly associated with upper-level westerlies that can be traced westward from the convective outflow in the central Pacific. The easterly outflow is weaker with subsidence near 95°E at the leading edge of the below-normal precipitation. Figure 13h also shows that the bulk of the troposphere over the Pacific Ocean and South America to be moist, consistent with the Kelvin wave decoupling seen in Fig. 10g.
The northeasterly surface wind anomalies over the Arabian Sea and the Bay of Bengal, seen in Fig. 13e, are associated with enhanced evaporation. In conjunction with the southerlies adjacent to the African coast there is low-level convergence (Figs. 13a–c) into the region of enhanced rainfall. The onset occurs in an easterly basic state, with the wind anomalies associated with the lower-tropospheric trailing anticyclones of the Rossby wave response to the suppressed convection. Importantly neither model I nor model II in Fig. 3 is applicable to the onset conditions.
A schematic of the onset conditions is given in Fig. 14. There are three main centers of enhanced convection, that over the western–central Pacific is the mature active phase of the MJO, and rainfall near western South America, possibly induced by the low pressure, convergence, and enhanced 1000-hPa specific humidity anomalies of the decoupled Kelvin wave. Last is the onset of the next active phase of the MJO in the western Indian Ocean. An open question is the extent to which circumnavigation of the aforementioned signals plays a role in the onset of the convection. At a minimum the East African highlands delay the eastward propagating signals, or they may impede the signal completely. For example, Fig. 12b shows that the free Kelvin wave convergence signal reaches the East African highlands on about day 13, while immediately to the east at ∼35°E divergence dominates from day 1 through 25. Thus, from day 1 onward there is no clear indication that the convergence signal from the west transitions over the highlands. Rather the convergence at ∼35°E, present from day −24 to 0, is found subsequently near 40°E. Additionally, the rainfall over Africa is below normal on day 10, only becoming positive on day 20, subsequent to enhanced 1000-hPa specific humidity (Fig. 10g) and the initial MJO convection in the Indian Ocean (Fig. 10a).
This dynamical framework is different from the “developing convection” conditions discussed in Fig. 1 of Hendon and Salby (1994) in which low-level convergence leads the convection by about 50° longitude. Their figure is consistent with our day 25 conditions, when the convection over the Indian Ocean and Africa has matured (not shown). Their truncation to zonal wavenumbers 1–3 was chosen “in favor of a single convective envelope across the eastern hemisphere.” The new scenario presented here indicates that the low-level moisture convergence paradigm is not operating during the onset phase of MJO convection.
c. The Indian Ocean versus Indonesia–west Pacific
EOF-2/PC-2 characterize the behavior over the Indian Ocean and the western–central Pacific Ocean (Fig. 2g) where the strongest convective signal is near 90°E. Other than a slight increase in amplitude of the anomalies over the Indian Ocean, the propagation characteristics given in Fig. 10 for PC-1 represent well the regressions obtained using PC-2 (taking into account the phase shift in time; not shown). This is due to the strong lag correlation between PC-2 and PC-1.
The vertical structure over the Indian Ocean is different from that over the western Pacific. This is related to the suppressed phase of the MJO near 150°E that is seen throughout most of the atmospheric column as anomalies of divergence, downward vertical velocity, and drying on day 0 for regression using PC-2 (Figs. 15a–d). Compared to the convection over Indonesia and the west Pacific (Fig. 7c), the free-tropospheric lateral inflow into the convective region is more asymmetrical, with the easterlies being twice as strong as the westerly inflow (Fig. 15c). Even so, a westward tilt is evident with low-level divergence and increased moisture to the east of the convection. The meridional structure at 90°E also exhibits differences relative to that at 125°E. Convergent flow only extends up to 400 hPa (Fig. 15e), though it, the vertical velocity, and specific humidity are more symmetrical about the equator (Figs. 15f–h versus Figs. 7f–h).
Lag–pressure plots of divergence, vertical velocity, winds, and specific humidity at 5°N–5°S, 90°E are shown in Fig. 16. As for the Pacific sector (Fig. 11), low-level moisture convergence over the Indian Ocean is important. From about day −25 to −12, anomalies of convergence and downward vertical velocity and below-normal moisture are evident through much of the troposphere during the suppressed phase of the MJO. This condition confines the low-level upward vertical velocity and convergence near the surface. As this competition becomes more favorable for convection, moistening of the lower troposphere precedes the development of convection at 90°E. Thus, once the convection is well established over the Indian Ocean, low-level moisture convergence is operative during MJO propagation.
4. Discussion and conclusions
In the methodology used here, years of strong MJO activity have been selected based on an index of the upper-tropospheric zonal mean zonal wind in conjunction with robust lead–lag relationships of eastward intraseasonal propagating OLR. Compared to a conventional analysis that uses successive years of data, the strong MJO years have an enhanced convective signal over Indonesia and the western Pacific, explain a greater fraction of local intraseasonal variability, and the phase relationship between the relevant principal component time series strengthens.
These strong MJO years are then used to describe the space–time structure of the MJO. The most pronounced finding is that the onset of MJO convection in the western Indian Ocean occurs in an easterly basic state and does not reflect the low-level moisture convergence paradigm. At this time, high pressure anomalies dominate the Indian Ocean, with lower pressure to the west, consistent with the low-level easterlies. The high pressure is associated with subsidence from the suppressed phase of the MJO, while the low pressure to the west is associated with the free Kelvin wave from the previous active phase of the MJO. The low-level convergence and specific humidity are in phase with the developing convection contrary to the low-level moisture convergence mechanism. However, once the convection has matured over the western Indian Ocean and the pressure gradient has reversed, low-level westerly anomalies become prevalent, and the low-level convergence, vertical velocity, and moisture lead the eastward propagating convection.
The degree to which circumnavigation of the decoupled Kelvin wave plays a role in the onset of the next phase of the MJO is an open question, particularly for the low-level divergence where the East African highlands impede the signal. One possible reason is that this is an artifact of the vertical extrapolation of the signal below ground. Alternatively, the impediment of the divergence signal may be real, which suggests that the highlands may play a role in preferentially confining the onset of MJO convection over the Indian Ocean.
The MJO has a rich vertical structure that reflects an important role for free-tropospheric interactions during the life cycle of the MJO. Over the Indian Ocean the system is more vertically stacked than over Indonesia and the western Pacific, though the westward vertical tilt is still evident. As the convection propagates southwestward into the SPCZ, the wind over the tropical eastern Pacific is no longer baroclinic, but rather westerlies prevail in the upper troposphere and near the surface, the latter being associated with the free Kelvin wave that decouples from the convection.
The dichotomy between the onset conditions and the mechanism of near-equatorial eastward propagation, along with the different off-equatorial behavior (low-level convergence in phase with the convection) attests to the complexity of the MJO. Having to represent such diverse interactions probably contributes to the difficulties general circulation models have in representing the MJO. This further supports the notion that simulation of the MJO is a critical test of a model's ability to represent the Tropics.
While this work has shed new light on the vertical structure of the MJO, and the conditions prevalent during the onset of the MJO convection, this applies to the situation when the MJO is in the process of making several circuits of the globe (as is typical during an active year). However, an understanding of the conditions that give rise to the first event during strong MJO years remains elusive and will be the subject of further investigation.
A caveat of this work is that many of the variables analyzed have been termed class B or C by Kalnay et al. (1996), in which the user is warned to be especially cautious in their interpretation. Validation of the surface fluxes indicates that the net shortwave is the most problematic, though overall the phasing of the surface fluxes is well captured relative to buoy data and other independent estimates. The amplitude of intraseasonal net surface heat flux and surface wind stress is well captured (Shinoda et al. 1999). Even so, the regressions herein, over a large number of events, underestimates the anomalies present during individual events, though compared to previous works the lead–lag relationships remain intact (e.g., Woolnough et al. 2000). This suggests that the comprehensive analysis presented here is at least schematically correct. However, an investigation using the forthcoming 40-yr reanalysis from the European Centre for Medium-Range Weather Forecasts will provide an alternative framework to confirm or deny these results, particularly with respect to the onset conditions of MJO convection.
Acknowledgments
I would like to thank Prof. Julia M. Slingo for useful comments on an earlier draft of this paper and for insightful discussions along with Drs. Pete Inness and Steve Woolnough (Reading University). Drs. Chidong Zhang (RSMS) and Harry Hendon (BMRC) clarified issues regarding model I and model II. Dr. George Kiladis (NOAA) provided helpful discussions regarding the regression technique. Dr. Krishna AchutaRao (PCMDI) provided assistance in processing the NCEP–NCAR reanalysis. The NCEP–NCAR reanalysis was obtained from the NOAA–CIRES Climate Diagnostics Center, Boulder, Colorado (http://www.cdc.noaa.gov/). I gratefully acknowledge the support of the NCAS Centre for Global Atmospheric Modelling and the Department of Meteorology at the University of Reading, United Kingdom, at which a portion of this work was completed. This work was performed under the auspices of the U.S. Department of Energy by the University of California Lawrence Livermore National Laboratory under Contract W-7405-Eng-48.
REFERENCES
Annamalai, H., and J. M. Slingo, 2001: Active/break cycles: Diagnosis of the intraseasonal variability of the Asian summer monsoon. Climate Dyn., 18 , 85–102.
Arkin, P. A., and P. E. Ardanuy, 1989: Estimating climatic-scale precipitation from space: A review. J. Climate, 2 , 1229–1238.
Bladé, I., and D. L. Hartmann, 1993: Tropical intraseasonal oscillations in a simple nonlinear model. J. Atmos. Sci., 50 , 2922–2939.
Emanuel, K. A., 1987: An air–sea interaction model of intraseasonal oscillations in the Tropics. J. Atmos. Sci., 44 , 2324–2340.
Ferranti, L., T. N. Palmer, F. Molteni, and E. Klinker, 1990: Tropical–extratropical interaction associated with the 30–60 day oscillation and its impact on medium and extended range prediction. J. Atmos. Sci., 47 , 2177–2199.
Flatau, M., P. J. Flatau, P. Phoebus, and P. P. Niiler, 1997: The feedback between equatorial convection and local radiative and evaporative processes: The implication for intraseasonal oscillations. J. Atmos. Sci., 54 , 2373–2386.
Gautier, C., G. Diak, and S. Masse, 1980: A simple physical model to estimate incident solar radiation at the surface from GOES satellite data. J. Appl. Meteor., 19 , 1005–1012.
Hendon, H. H., and M. L. Salby, 1994: The life cycle of the Madden–Julian oscillation. J. Atmos. Sci., 51 , 2225–2237.
Hendon, H. H., C. Zhang, and J. D. Glick, 1999: Interannual variation of the Madden–Julian oscillation during austral summer. J. Climate, 12 , 2538–2550.
Hosom, D. S., R. A. Weller, R. E. Payne, and K. E. Prada, 1995: The IMET (improved meteorology) ship and buoy systems. J. Atmos. Oceanic Technol., 12 , 527–540.
Jones, C., and B. C. Weare, 1996: The role of low-level moisture convergence and ocean latent heat fluxes in the Madden and Julian oscillation: An observational analysis using ISCCP data and ECMWF analyses. J. Climate, 9 , 3086–3140.
Jones, C., D. E. Waliser, and C. Gautier, 1998: The influence of the Madden–Julian oscillation on ocean surface heat fluxes and sea surface temperature. J. Climate, 11 , 1057–1072.
Kalnay, E., and Coauthors. 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Kemball-Cook, S. R., and B. C. Weare, 2001: The onset of convection in the Madden–Julian oscillation. J. Climate, 14 , 780–793.
Knutson, T. R., and K. M. Weickmann, 1987: 30–60 day atmospheric oscillations: Composite life cycles of convection and circulation anomalies. Mon. Wea. Rev., 115 , 1407–1436.
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.
Lau, K-M., and C. H. Sui, 1997: Mechanisms of short-term sea surface temperature regulation: Observations during TOGA-COARE. J. Climate, 10 , 465–472.
Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) OLR dataset. Bull. Amer. Meteor. Soc., 77 , 1275–1277.
Lin, X., and R. H. Johnson, 1996: Heating, moistening, and rainfall over the western Pacific warm pool during TOGA COARE. J. Atmos. Sci., 53 , 3367–3383.
Livezey, R. E., and W. Y. Chen, 1983: Statistical field significance and its determination by Monte Carlo techniques. Mon. Wea. Rev., 111 , 46–59.
Madden, R. A., 1988: Large intraseasonal variations in wind stress over the tropical Pacific. J. Geophys. Res., 93 , 5333–5340.
Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28 , 702–708.
Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the Tropics with a 40–50 day period. J. Atmos. Sci., 29 , 1109–1123.
Madden, R. A., and P. R. Julian, 1994: Observations of the 40–50-day tropical oscillation—A review. Mon. Wea. Rev., 122 , 814–837.
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.
Matthews, A. J., 2000: Propagation mechanisms for the Madden–Julian oscillation. Quart. J. Roy. Meteor. Soc., 126 , 2637–2651.
Mo, K. C., and R. W. Higgins, 1996: Large-scale atmospheric moisture transport as evaluated in the NCEP/NCAR and the NASA/DAO reanalyses. J. Climate, 9 , 1531–1545.
Mo, K. C., and R. W. Higgins, 1998: Tropical influences on California precipitation. J. Climate, 11 , 412–430.
Murakami, T., 1988: Intraseasonal atmospheric teleconnection patterns during Northern Hemisphere winter. J. Climate, 1 , 117–131.
Myers, D. S., and D. E. Waliser, 2003: Three-dimensional water vapor and cloud variations associated with the Madden–Julian oscillation during Northern Hemisphere winter. J. Climate, 16 , 929–950.
Neelin, J. D., I. M. Held, and K. H. Cook, 1987: Evaporation wind feedback and low-frequency variability in the tropical atmosphere. J. Atmos. Sci., 44 , 2341–2348.
Nitta, T., T. Mizuno, and K. Takahashi, 1992: Multi-scaled convective systems and the 86/87 ENSO. J. Meteor. Soc. Japan, 70 , 447–458.
Pan, H. L., and L. Mahrt, 1987: Interaction between soil hydrology and boundary layer developments. Bound.-Layer Meteor., 38 , 185–202.
Pan, H. L., and W. S. Wu, 1994: Implementing a mass-flux convection parameterization package for the NMC medium-range forecast model. Preprints, 10th Conf. on Numerical Weather Prediction, Portland, OR, Amer. Meteor. Soc., 96–98.
Reynolds, R. W., and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7 , 929–948.
Rosen, R. D., and D. A. Salstein, 1983: Variations in angular momentum on global and regional scales and the length of day. J. Geophys. Res., 88 , 5451–5470.
Rossow, W. B., L. C. Garder, P. J. Lu, and A. W. Walker, 1988: International Satellite Cloud Climatology Project (ISSCP) documentation of cloud data. WMO Tech. Doc. WMO/TD 266, World Meteorological Organization, Geneva, Switzerland, 77 pp.
Rui, H., and B. Wang, 1990: Development characteristics and dynamic structure of tropical intraseasonal convective anomalies. J. Atmos. Sci., 47 , 357–379.
Salby, M. L., and H. H. Hendon, 1994: Intraseasonal behavior of clouds, temperature, and motion in the Tropics. J. Atmos. Sci., 51 , 2207–2224.
Salby, M. L., R. Garcia, and H. H. Hendon, 1994: Planetary-scale circulations in the presence of climatological and wave-induced heating. J. Atmos. Sci., 51 , 2344–2367.
Shinoda, T., H. H. Hendon, and J. Glick, 1998: Intraseasonal variability of surface fluxes and sea surface temperature in the tropical western Pacific and Indian oceans. J. Climate, 11 , 1685–1702.
Shinoda, T., H. H. Hendon, and J. Glick, 1999: Intraseasonal surface fluxes in the tropical western Pacific and Indian Oceans from NCEP reanalyses. Mon. Wea. Rev., 127 , 678–693.
Slingo, J. M., 1987: The development and verification of a cloud prediction scheme for the ECMWF model. Quart. J. Roy. Meteor. Soc., 113 , 899–927.
Slingo, J. M., and Coauthors. 1996: Intraseasonal oscillations in 15 atmospheric general circulation models: Results from an AMIP diagnostic subproject. Climate Dyn., 12 , 325–357.
Slingo, J. M., D. P. Rowell, K. R. Sperber, and F. Nortley, 1999: On the predictability of the interannual behaviour of the Madden–Julian oscillation and its relationship with El Niño. Quart. J. Roy. Meteor. Soc., 125 , 583–609.
Sperber, K. R., J. M. Slingo, P. M. Inness, and W. K-M. Lau, 1997: On the maintenance and initiation of the intraseasonal oscillation in the NCEP/NCAR reanalysis and in the GLA and UKMO AMIP simulations. Climate Dyn., 13 , 769–795.
Stendel, M., and K. Arpe, 1997: Evaluation of the hydrological cycle in reanalyses and observations. Rep. 228, Max-Planck-Institut für Meteorologie, Hamburg, Germany, 52 pp.
Wang, B., and T. Li, 1994: Convective interaction with boundary-layer dynamics in the development of a tropical intraseasonal system. J. Atmos. Sci., 51 , 1386–1400.
Wang, B., and X. Xie, 1998: Coupled modes of the warm pool climate system. Part I: The role of air–sea interaction in maintaining the Madden–Julian oscillation. J. Climate, 11 , 2116–2135.
Weickmann, K. M., G. R. Lussky, and J. E. Kutzbach, 1985: Intraseasonal (30–60 day) fluctuations of outgoing longwave radiation and 250 mb streamfunction during northern winter. Mon. Wea. Rev., 113 , 941–961.
Woolnough, S. J., J. M. Slingo, and B. J. Hoskins, 2000: The relationship between convection and sea surface temperature on intraseasonal timescales. J. Climate, 13 , 2086–2104.
Woolnough, S. J., J. M. Slingo, and B. J. Hoskins, 2001: The organization of tropical convection by intraseasonal sea surface temperature anomalies. Quart. J. Roy. Meteor. Soc., 127 , 887–907.
Xie, P., and P. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc., 78 , 2539–2558.
Zhang, C., 1996: Atmospheric intraseasonal variability at the surface in the tropical western Pacific Ocean. J. Atmos. Sci., 53 , 739–758.
Zhang, C., and M. J. McPhaden, 1995: The relationship between sea surface temperature and latent heat flux in the equatorial Pacific. J. Climate, 8 , 589–605.
Zhang, C., and M. J. McPhaden, 2000: Intraseasonal surface cooling in the equatorial western Pacific. J. Climate, 13 , 2261–2276.
Variance in a 101-day moving window of 20–100-day bandpass-filtered 200-hPa zonal mean zonal wind averaged between 10°N and 10°S. Above the peaks in variance the numbers are max correlations and L “n” is the time lag in days at which it occurred based on the two leading PCs of 20–100-day bandpass-filtered AVHRR OLR, which characterize the eastward propagation of MJO convection. The values in red are based on the analysis of 7 yr of strong MJO activity, corresponding to the PCs in Fig. 2i, and those in blue are when 10 strong years were analyzed (see footnote 1)
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Variance in a 101-day moving window of 20–100-day bandpass-filtered 200-hPa zonal mean zonal wind averaged between 10°N and 10°S. Above the peaks in variance the numbers are max correlations and L “n” is the time lag in days at which it occurred based on the two leading PCs of 20–100-day bandpass-filtered AVHRR OLR, which characterize the eastward propagation of MJO convection. The values in red are based on the analysis of 7 yr of strong MJO activity, corresponding to the PCs in Fig. 2i, and those in blue are when 10 strong years were analyzed (see footnote 1)
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Variance in a 101-day moving window of 20–100-day bandpass-filtered 200-hPa zonal mean zonal wind averaged between 10°N and 10°S. Above the peaks in variance the numbers are max correlations and L “n” is the time lag in days at which it occurred based on the two leading PCs of 20–100-day bandpass-filtered AVHRR OLR, which characterize the eastward propagation of MJO convection. The values in red are based on the analysis of 7 yr of strong MJO activity, corresponding to the PCs in Fig. 2i, and those in blue are when 10 strong years were analyzed (see footnote 1)
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Analysis of AVHRR OLR for Nov–Mar. (a) EOF-1 of 20–100-day bandpass-filtered OLR for 1982–83–1990–91. The EOF has been scaled by a 1-std-dev perturbation of PC-1, giving units of W m−2. (b) As in (a), but for EOF-2. (c) Local percent variance explained for EOF-1 + EOF-2. (d) Principal component time series of EOF-1 and EOF-2. (e) Fast Fourier transform power spectra for PC-1 and PC-2. (f)–(j) As in (a)–(c), but for the 7 yr of strong MJO activity shown in Fig. 1
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Analysis of AVHRR OLR for Nov–Mar. (a) EOF-1 of 20–100-day bandpass-filtered OLR for 1982–83–1990–91. The EOF has been scaled by a 1-std-dev perturbation of PC-1, giving units of W m−2. (b) As in (a), but for EOF-2. (c) Local percent variance explained for EOF-1 + EOF-2. (d) Principal component time series of EOF-1 and EOF-2. (e) Fast Fourier transform power spectra for PC-1 and PC-2. (f)–(j) As in (a)–(c), but for the 7 yr of strong MJO activity shown in Fig. 1
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Analysis of AVHRR OLR for Nov–Mar. (a) EOF-1 of 20–100-day bandpass-filtered OLR for 1982–83–1990–91. The EOF has been scaled by a 1-std-dev perturbation of PC-1, giving units of W m−2. (b) As in (a), but for EOF-2. (c) Local percent variance explained for EOF-1 + EOF-2. (d) Principal component time series of EOF-1 and EOF-2. (e) Fast Fourier transform power spectra for PC-1 and PC-2. (f)–(j) As in (a)–(c), but for the 7 yr of strong MJO activity shown in Fig. 1
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Conceptual models of the relationship between surface fluxes and MJO convection (after Zhang and McPhaden 2000), where u is the zonal wind, τ is the zonal windstress, Q1 is the latent heat flux, Qs is the sensible the flux, Qsw is the shortwave flux, and P − E is precipitation minus evaporation
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Conceptual models of the relationship between surface fluxes and MJO convection (after Zhang and McPhaden 2000), where u is the zonal wind, τ is the zonal windstress, Q1 is the latent heat flux, Qs is the sensible the flux, Qsw is the shortwave flux, and P − E is precipitation minus evaporation
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Conceptual models of the relationship between surface fluxes and MJO convection (after Zhang and McPhaden 2000), where u is the zonal wind, τ is the zonal windstress, Q1 is the latent heat flux, Qs is the sensible the flux, Qsw is the shortwave flux, and P − E is precipitation minus evaporation
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Zero-time-lag linear regressions of PC-1 with 20–100-day bandpass-filtered (a) AVHRR OLR (W m−2) and 200-hPa wind (m s−1), (b) CMAP rainfall (mm day−1) and 850-hPa wind (m s−1), (c) 500-hPa vertical velocity (Pa s−1; positive downward), (d) net surface shortwave radiation (W m−2; positive downward), (e) SST (K), ground temperature (K), and surface winds (m s−1), (f) latent heat flux (W m−2; positive upward corresponds to evaporative cooling of the surface) and surface windstress (N m−2), (g) net surface heat flux (W m−2; positive downward), and (h) percent contribution of the latent heat flux to the net surface heat flux (the latent heat flux has been multiplied by −1 for this calculation). Wind vectors are plotted at every other grid point for clarity of presentation. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Zero-time-lag linear regressions of PC-1 with 20–100-day bandpass-filtered (a) AVHRR OLR (W m−2) and 200-hPa wind (m s−1), (b) CMAP rainfall (mm day−1) and 850-hPa wind (m s−1), (c) 500-hPa vertical velocity (Pa s−1; positive downward), (d) net surface shortwave radiation (W m−2; positive downward), (e) SST (K), ground temperature (K), and surface winds (m s−1), (f) latent heat flux (W m−2; positive upward corresponds to evaporative cooling of the surface) and surface windstress (N m−2), (g) net surface heat flux (W m−2; positive downward), and (h) percent contribution of the latent heat flux to the net surface heat flux (the latent heat flux has been multiplied by −1 for this calculation). Wind vectors are plotted at every other grid point for clarity of presentation. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Zero-time-lag linear regressions of PC-1 with 20–100-day bandpass-filtered (a) AVHRR OLR (W m−2) and 200-hPa wind (m s−1), (b) CMAP rainfall (mm day−1) and 850-hPa wind (m s−1), (c) 500-hPa vertical velocity (Pa s−1; positive downward), (d) net surface shortwave radiation (W m−2; positive downward), (e) SST (K), ground temperature (K), and surface winds (m s−1), (f) latent heat flux (W m−2; positive upward corresponds to evaporative cooling of the surface) and surface windstress (N m−2), (g) net surface heat flux (W m−2; positive downward), and (h) percent contribution of the latent heat flux to the net surface heat flux (the latent heat flux has been multiplied by −1 for this calculation). Wind vectors are plotted at every other grid point for clarity of presentation. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Figs. 4a–g, but for unfiltered data. In Fig. 5e the units of SST are °C. The same number of degrees of freedom have been used for both the filtered and unfiltered data. Thus, given the greater variance in the unfiltered data, fewer grid points exhibit statistical significance relative to the filtered data in Fig. 4
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Figs. 4a–g, but for unfiltered data. In Fig. 5e the units of SST are °C. The same number of degrees of freedom have been used for both the filtered and unfiltered data. Thus, given the greater variance in the unfiltered data, fewer grid points exhibit statistical significance relative to the filtered data in Fig. 4
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Figs. 4a–g, but for unfiltered data. In Fig. 5e the units of SST are °C. The same number of degrees of freedom have been used for both the filtered and unfiltered data. Thus, given the greater variance in the unfiltered data, fewer grid points exhibit statistical significance relative to the filtered data in Fig. 4
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 4, but for 20–100-day bandpass-filtered (a) 1000-hPa specific humidity (kg kg−1) and (b) 1000-hPa divergence (s−1).
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 4, but for 20–100-day bandpass-filtered (a) 1000-hPa specific humidity (kg kg−1) and (b) 1000-hPa divergence (s−1).
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 4, but for 20–100-day bandpass-filtered (a) 1000-hPa specific humidity (kg kg−1) and (b) 1000-hPa divergence (s−1).
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Longitude–height cross sections of zero-time-lag linear regressions of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered (a) divergence (s−1), (b) vertical velocity (Pa s−1), (c) u wind and vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the u wind (m s−1)] and contours of the u wind in increments of 0.5 m s−1, (d) specific humidity (kg kg−1). In (a)–(d) the vertical dashed line at 125°E is the longitude of strongest convection. Latitude–height cross sections at 125°E of zero-time-lag linear regressions of PC-1 with 20–100-day bandpass-filtered (e) divergence (s−1), (f) vertical velocity (Pa s−1), (g) υ wind–vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the υ wind (m s−1)] and contours of the υ wind in increments of 0.1 m s−1, (h) specific humidity (kg kg−1). In (c) wind vectors have been plotted at every other longitude. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Longitude–height cross sections of zero-time-lag linear regressions of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered (a) divergence (s−1), (b) vertical velocity (Pa s−1), (c) u wind and vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the u wind (m s−1)] and contours of the u wind in increments of 0.5 m s−1, (d) specific humidity (kg kg−1). In (a)–(d) the vertical dashed line at 125°E is the longitude of strongest convection. Latitude–height cross sections at 125°E of zero-time-lag linear regressions of PC-1 with 20–100-day bandpass-filtered (e) divergence (s−1), (f) vertical velocity (Pa s−1), (g) υ wind–vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the υ wind (m s−1)] and contours of the υ wind in increments of 0.1 m s−1, (h) specific humidity (kg kg−1). In (c) wind vectors have been plotted at every other longitude. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Longitude–height cross sections of zero-time-lag linear regressions of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered (a) divergence (s−1), (b) vertical velocity (Pa s−1), (c) u wind and vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the u wind (m s−1)] and contours of the u wind in increments of 0.5 m s−1, (d) specific humidity (kg kg−1). In (a)–(d) the vertical dashed line at 125°E is the longitude of strongest convection. Latitude–height cross sections at 125°E of zero-time-lag linear regressions of PC-1 with 20–100-day bandpass-filtered (e) divergence (s−1), (f) vertical velocity (Pa s−1), (g) υ wind–vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the υ wind (m s−1)] and contours of the υ wind in increments of 0.1 m s−1, (h) specific humidity (kg kg−1). In (c) wind vectors have been plotted at every other longitude. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Linear regressions of PC-1 with 20–100-day bandpass-filtered CMAP rainfall (mm day−1) and 850-hPa wind (m s−1) at time lags of (a) −20, (b) −10, (c) 0, (d) 10, and (e) 20 days. (f)–(j) As in (a)–(e), but for linear regressions of PC-1 with 20–100-day bandpass-filtered net surface heat flux (W m−2) and zonal wind stress (N m−2). Wind vectors are plotted at every other grid point. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Linear regressions of PC-1 with 20–100-day bandpass-filtered CMAP rainfall (mm day−1) and 850-hPa wind (m s−1) at time lags of (a) −20, (b) −10, (c) 0, (d) 10, and (e) 20 days. (f)–(j) As in (a)–(e), but for linear regressions of PC-1 with 20–100-day bandpass-filtered net surface heat flux (W m−2) and zonal wind stress (N m−2). Wind vectors are plotted at every other grid point. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Linear regressions of PC-1 with 20–100-day bandpass-filtered CMAP rainfall (mm day−1) and 850-hPa wind (m s−1) at time lags of (a) −20, (b) −10, (c) 0, (d) 10, and (e) 20 days. (f)–(j) As in (a)–(e), but for linear regressions of PC-1 with 20–100-day bandpass-filtered net surface heat flux (W m−2) and zonal wind stress (N m−2). Wind vectors are plotted at every other grid point. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 8, but for 20–100-day bandpass-filtered SST and ground temperature (K) and surface wind (m s−1).
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 8, but for 20–100-day bandpass-filtered SST and ground temperature (K) and surface wind (m s−1).
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 8, but for 20–100-day bandpass-filtered SST and ground temperature (K) and surface wind (m s−1).
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Longitude–time lag plots of the linear regression of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered (a) CMAP rainfall (mm day−1), (b) 500-hPa vertical velocity (Pa s−1; positive downward), (c) net surface shortwave radiation (W m−2; positive downward), (d) zonal wind stress (N m−2), (e) latent heat flux (W m−2; positive upward corresponds to evaporative cooling of the surface), (f) SST and surface temperature (K), (g) 1000-hPa specific humidity (kg kg−1), (h) 1000-hPa divergence (s−1). On each plot contours of the regression of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered OLR are plotted in increments of 2.5 W m−2. Time lags run from −25 to 25 days. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units. The vertical dashed line gives the central longitude of convection in EOF-1 (Fig. 2a), and the horizontal dashed line corresponds to zero time lag
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Longitude–time lag plots of the linear regression of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered (a) CMAP rainfall (mm day−1), (b) 500-hPa vertical velocity (Pa s−1; positive downward), (c) net surface shortwave radiation (W m−2; positive downward), (d) zonal wind stress (N m−2), (e) latent heat flux (W m−2; positive upward corresponds to evaporative cooling of the surface), (f) SST and surface temperature (K), (g) 1000-hPa specific humidity (kg kg−1), (h) 1000-hPa divergence (s−1). On each plot contours of the regression of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered OLR are plotted in increments of 2.5 W m−2. Time lags run from −25 to 25 days. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units. The vertical dashed line gives the central longitude of convection in EOF-1 (Fig. 2a), and the horizontal dashed line corresponds to zero time lag
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Longitude–time lag plots of the linear regression of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered (a) CMAP rainfall (mm day−1), (b) 500-hPa vertical velocity (Pa s−1; positive downward), (c) net surface shortwave radiation (W m−2; positive downward), (d) zonal wind stress (N m−2), (e) latent heat flux (W m−2; positive upward corresponds to evaporative cooling of the surface), (f) SST and surface temperature (K), (g) 1000-hPa specific humidity (kg kg−1), (h) 1000-hPa divergence (s−1). On each plot contours of the regression of PC-1 with 5°N–5°S-averaged 20–100-day bandpass-filtered OLR are plotted in increments of 2.5 W m−2. Time lags run from −25 to 25 days. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units. The vertical dashed line gives the central longitude of convection in EOF-1 (Fig. 2a), and the horizontal dashed line corresponds to zero time lag
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Time lag vs height plots of linear regressions of PC-1 with 125°E (5°N–5°S averaged) 20–100-day bandpass-filtered (a) divergence (s−1), (b) vertical velocity (Pa s−1; positive downward), (c) u wind and vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the u wind (m s−1)] and contours of the u wind in increments of 0.5 m s−1, and (d) specific humidity (kg kg−1). Time lags run from −25 to 25 days. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Time lag vs height plots of linear regressions of PC-1 with 125°E (5°N–5°S averaged) 20–100-day bandpass-filtered (a) divergence (s−1), (b) vertical velocity (Pa s−1; positive downward), (c) u wind and vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the u wind (m s−1)] and contours of the u wind in increments of 0.5 m s−1, and (d) specific humidity (kg kg−1). Time lags run from −25 to 25 days. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Time lag vs height plots of linear regressions of PC-1 with 125°E (5°N–5°S averaged) 20–100-day bandpass-filtered (a) divergence (s−1), (b) vertical velocity (Pa s−1; positive downward), (c) u wind and vertical velocity vectors [N. B., the vertical velocity (Pa s−1) has been multiplied by −100 to give scaling compatible with the u wind (m s−1)] and contours of the u wind in increments of 0.5 m s−1, and (d) specific humidity (kg kg−1). Time lags run from −25 to 25 days. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 10, but for 20–100-day bandpass-filtered (a) SLP (hPa) and contours of OLR in increments of 2.5 W m−2, (b) 1000-hPa divergence (s−1) and contours of SLP in increments of 0.1 hPa, (c) zonal wind stress (N m−2) and contours of SLP in increments of 0.1 hPa. In (c) the arrows highlight the eastward propagating low pressure surge impacting the Andes and the westward propagation of westerly zonal wind stress over the eastern Pacific
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 10, but for 20–100-day bandpass-filtered (a) SLP (hPa) and contours of OLR in increments of 2.5 W m−2, (b) 1000-hPa divergence (s−1) and contours of SLP in increments of 0.1 hPa, (c) zonal wind stress (N m−2) and contours of SLP in increments of 0.1 hPa. In (c) the arrows highlight the eastward propagating low pressure surge impacting the Andes and the westward propagation of westerly zonal wind stress over the eastern Pacific
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 10, but for 20–100-day bandpass-filtered (a) SLP (hPa) and contours of OLR in increments of 2.5 W m−2, (b) 1000-hPa divergence (s−1) and contours of SLP in increments of 0.1 hPa, (c) zonal wind stress (N m−2) and contours of SLP in increments of 0.1 hPa. In (c) the arrows highlight the eastward propagating low pressure surge impacting the Andes and the westward propagation of westerly zonal wind stress over the eastern Pacific
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Day 10 lag linear regressions of PC-1 with 20–100-day bandpass-filtered (a) 1000-hPa divergence (s−1), (b) 925-hPa divergence (s−1), (c) 850-hPa divergence (s−1), (d) longitude–height cross section of divergence (s−1), (e) latent heat flux (W m−2) and wind stress (N m−2), (f) 925-hPa specific humidity (kg kg−1) and wind (m s−1), (g) 850-hPa specific humidity (kg kg−1) and wind (m s−1), and (h) longitude–height cross section of specific humidity (kg kg−1) and vectors of zonal wind (m s−1) and vertical velocity (Pa s−1; −100×). In (e)–(g) wind vectors have been plotted at every other grid point. In (h) wind vectors have been plotted at every other longitude. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Day 10 lag linear regressions of PC-1 with 20–100-day bandpass-filtered (a) 1000-hPa divergence (s−1), (b) 925-hPa divergence (s−1), (c) 850-hPa divergence (s−1), (d) longitude–height cross section of divergence (s−1), (e) latent heat flux (W m−2) and wind stress (N m−2), (f) 925-hPa specific humidity (kg kg−1) and wind (m s−1), (g) 850-hPa specific humidity (kg kg−1) and wind (m s−1), and (h) longitude–height cross section of specific humidity (kg kg−1) and vectors of zonal wind (m s−1) and vertical velocity (Pa s−1; −100×). In (e)–(g) wind vectors have been plotted at every other grid point. In (h) wind vectors have been plotted at every other longitude. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Day 10 lag linear regressions of PC-1 with 20–100-day bandpass-filtered (a) 1000-hPa divergence (s−1), (b) 925-hPa divergence (s−1), (c) 850-hPa divergence (s−1), (d) longitude–height cross section of divergence (s−1), (e) latent heat flux (W m−2) and wind stress (N m−2), (f) 925-hPa specific humidity (kg kg−1) and wind (m s−1), (g) 850-hPa specific humidity (kg kg−1) and wind (m s−1), and (h) longitude–height cross section of specific humidity (kg kg−1) and vectors of zonal wind (m s−1) and vertical velocity (Pa s−1; −100×). In (e)–(g) wind vectors have been plotted at every other grid point. In (h) wind vectors have been plotted at every other longitude. All regressions have been scaled by a 1-std-dev perturbation of PC-1 to give the aforementioned units
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Schematic of the near-equatorial onset of MJO convection in the Indian Ocean on day 10 with respect to PC-1. This schematic is based on Figs. 8d, 9d, and 13e–h
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Schematic of the near-equatorial onset of MJO convection in the Indian Ocean on day 10 with respect to PC-1. This schematic is based on Figs. 8d, 9d, and 13e–h
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
Schematic of the near-equatorial onset of MJO convection in the Indian Ocean on day 10 with respect to PC-1. This schematic is based on Figs. 8d, 9d, and 13e–h
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 7, but for regressions using PC-2.
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 7, but for regressions using PC-2.
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 7, but for regressions using PC-2.
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 11, but for regressions at 90°E using PC-2.
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 11, but for regressions at 90°E using PC-2.
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
As in Fig. 11, but for regressions at 90°E using PC-2.
Citation: Monthly Weather Review 131, 12; 10.1175/1520-0493(2003)131<3018:PATVSO>2.0.CO;2
In Fig. 1, 10 yr of strong MJO activity were initially identified. An EOF analysis using the 10 strong years gave similar patterns as those for the seven strong years in Figs. 2f–j, but with reduced variance explained, and minor changes in the centers of action and the PCs. For these 10 yr the maximum positive correlation and the time lag at which it occurred displayed a wide range of values during the individual years. With the exclusion of 1979–80, 1988–89, and 1998–99 (which had their maximum positive correlation at time lags greater than or equal to 20 days) the maximum correlation increased during five of the seven remaining years (slightly decreasing in 1991–92). Furthermore, the time lag at which the maximum correlation occurred came into more uniform agreement, ranging from 11 to 14 days. This has the benefit of minimizing phase differences between events that might otherwise obscure the regressions. Note that boreal summer maxima were not included in the analysis (e.g., 1981). Annamalai and Slingo (2001) evaluated the eastward and northward propagating components of summer monsoon intraseasonal variability.
During TOGA COARE three strong MJO events occurred, though this is not apparent from the MJO index in Fig. 1. During this period the AVHRR OLR projected strongly onto EOFs 1 and 2 (Figs. 2f and 2g), resulting in a maximum positive correlation of 0.79 at an 8-day time lag between PC-2 and PC-1. As such, this year would have been excluded from the analysis in an effort to minimize the phase differences between events (see section 3a for more details). Though the TOGA COARE events tended to have a faster propagation rate on average, the relationship between the convection and the surface fluxes in the results presented herein are consistent with observations taken during the intensive observation period (e.g., Lau and Sui 1997).
