The Madden-Julian Oscillation:
A coupled atmopshere-ocean phenomenon?

1. Introduction
2. The oceanic response to convection on intraseasonal timescales
3. The organisation of convection by intraseasonal SST anomalies
4. Implications for the MJO

1. Introduction.

1.1 Background.

The Madden-Julian Oscillation (MJO), or intraseasonal oscillation is the dominant mode of subseasonal variability in the tropics (Madden and Julian 1971,1972,1994). It is manifested as large-scale eastward propagating circulation anomalies and associated convective anomalies with timesclaes of about 30-60 days. The convective anomalies associated with the intraseasonal oscillation are strongest over the Indian Ocean, Maritime Continent and warm pool region of the west Pacific. Over the cooler waters of the eastern and central Pacific and the Atlantic Ocean, there is very little convective signature. The oscillation is strongest and most frequent in the boreal winter and spring. Over the Eastern Hemisphere were the convective signal is strong, the MJO has a phase speed of around 5 ms-1. Away from the convective signal the phase speed is about 10ms-1 (Hendon and Salby 1994). There is a considerable amount of variability in the activity of the MJO, in terms of both the strength of the anomalies and the number of events in the season (Slingo et al. 1999)

1.2 Theories and Models of the MJO.

Theories and Models of the MJO need to be able to represent and predict the observed nature of the MJO, in terms of its propagation characteristics, seasonality and spatial extent of the convective anomalies. Comparisons of the MJO in general circulation models as part of AMIP (Slingo et al. 1996) indicated that most GCMs had weak intraseasonal activity; they tended to simulate slightly shorter periods than observations and failed to capture the seasonality of the MJO.

There have been many attempts to develop a theoretical framework to explain the propagation characteristics of the MJO. Much of this workhas centred around the modification of equatorially trapped waves by considering the effect of moist processes, either through the wave-CISK mechanism (e.g. Lau and Peng, 1987) or by considering the role of evaporation-wind feedbacks (Emmanuel 1987; Neelin et al 1987). Both of these theories have difficulty predicting the correct phase speed and spatial scale for the oscillation, but most importantly, the evaportation-wind feedback mechanism requires basic state easterlies at the surface. In the region where the convective signal is acitve the climatological winds are westerly at the surface.

1.3 Does the ocean play a role in the Madden-Julian Oscillation?

Observations from the Tropical Ocean Global Atmosphere Coupled Ocean Atmosphere Response Experiment (TOGA COARE) have shown that SSTs in the warm pool are modulated by the passage of the MJO (e.g Weller and Anderson 1996; Hendon and Glick 1997; Lau and Sui 1997). The modulation of the SST is related to the change in surface fluxes associated with the passage of the MJO.

Such observations have led to speculation that the MJO may be a coupled mode of the atmosphere-ocean system (e.g. Flatau et al 1997; Sperber et al 1997). Model experiments by Flatau et al (1997) and Waliser et al (1999) have shown that incorporating a coupling between the convection and SST can enhance the models representation of the MJO. Wang and Xie (1998) have developed a simple linear model of the tropical atmosphere system which, for parameters consistent with the observed values for the warm pool, exhibits unstable eastward propagating modes with phase speeds under 10ms-1. The large scale features of this mode compare well with the structure of the MJO as observed during TOGA COARE.

The hypothesis that the MJO is a coupled ocean-atmosphere phenomenon requires that;

  1. The atmospheric MJO impacts on the ocean.
  2. Intraseasonal variations in SST can organise convection.
These two issues will be addressed individually, the first through the analysis of observational data and the second through aquaplanet GCM integrations.

2. The oceanic response to convection on intraseasonal timescales.

(Woolnough et al. 2000a)

2.1 Data.

The response of the ocean to intraseasonal variations in convection has been examined using 15 years of data from 1982-1997 inclusive. Outgoing longwave radiation from the NOAA polar orbiting satellites is used as a proxy for convection. Surface fluxes of latent heat and shortwave radiation and the surface wind stresses were obtained from the ECMWF reanalysis and operational analyses. The Reynold's weekly SST data were interpolated onto a daily timeseries for consistency with the OLR and surface fluxes. To isolate the intraseasonal variability a 20-100day bandpass filter was applied. In view of the very different nature of the tropical circulation and the behaviour of the MJO during the boreal summer only the period of October to May is analysed.

Figure 1: 20-100days Bandpass anomalies of OLR, SST, shortwave flux, latent heat flux and zonal wind stress, for 01 Oct 97 - 01 JUN 88, averaged between 5°N and 5°S. Negative OLR anomalies correspond to enhanced convetion, surface flux anomalies are postive into the surface.
In figure 1 several eastward propagating OLR anomalies can be seen, associated with these OLR anomalies there are varitations in the SST, shortwave flux (SWF), latent heat flux and zonal wind stress (UST).

2.2 Lag-correlations

To investigate whether the relationships between the convection and surface fields are coherent lag-correlations between the equatorially averaged OLR and SST, SWF, LHF, and UST were calculated. Figure 2 shows the lag-correlation maps for the 15 seasons treated as a single time series.

Figure 2: Lag-correlations of OLR with SST, SWF, LHF and UST. Contours every 0.1, red indicates correlations > 0.1, blue indicates correlations < 0.1. Positive lags correspond to OLR anomalies preceding anomalies in the surface fields. The sign convection is such that postive correlations correspond to enhanced convection (negative OLR anomaly) being correlated with negative SST anomalies, reduced SWF, westerly wind dtress anomalies and enhanced evaporation.
The lag-correlation maps indicate that enhanced convection associated with reduction in the shortwave flux at the surface as expected. Between 60°E and 180° positve SST anomalies lead enhanced convection by about 10 days and negative SST anomalies follow enhanced convection by about 10 days. Also extending out in the central Pacific enhanced convection is preceded by easterly wind stress anomalies and followed by westerly windstress anomalies. In the region where the climatological winds are westerly at the surface these windstress anomalies corresopond to reduced evaporation and surface warming prior to the convection and enhanced evaporation and surface cooling following the convection. To the east of the dateline where the surface winds becomed easterly this relationship betwen the convection and the latent heat flux breaks down.

To investigate the robustness of these relationships from year to year the lag-correlations where calculated for each season individually and for 3 El Niño years and 3 La Niña years. In El Niño years the region over which the relationship between the convection, the SST and the LHF holds extends further beyond the dateline, related to the extension of the warm pool and westerly winds into the central Pacific during those years. Similarly in La Niña years the region over which these relationships hold is more confined.

Year to year variations are in these relationships are more difficult to summarize but can be used to provide a measure of the robustness of the mean relationship between the convection and the surface fields. Figure 3 shows a summary of the relationship between the convection and the surface fields based on the lag at which the maxima and mininma in the lag-correlations occur, the points are determined from the single timeseries correlation coefficients and the error bars from the spread of lags from the individual seasons (see Woolnough et al. (2000a) for the details).

Figure 3: Summary of the temporal relationships between the convection and surface fields. The dots indicate the lags at which extrema in correlation coefficients occur when the 15 seasons are treated as one time series. Within each region the ordering of the events (relative to a convective maximum ) from left to right is SWF max, UST min, LHF max, SST max, SWF min, UST max, LHF min, SST min, SWF max. Here SWF and LHF positive corresponds to fluxes into the surface. (see Woolnough et al. (2000a) for an explaination of the error bars)
The timing of each event relative to the convective maximum is consistent from one region to another; High SWF, easterly wind stress anomalies (giving light total winds), and reduced evaporation lead the positive SST anomalies, consistent with the surface fluxes generating the SST anomalies. Following the warm SST anomalies there is enhanced convection and reduced SWF, and a few days later, enhanced westerly winds and strong evaporation leading to negative SST anomalies. This relationship is most robust in the Indian Ocean and West Pacific, but holds less well over the Maritime Continent region where the surface processes may be disrupted by the presence of the land. There is some interannual variability in the timing of these events, but the general relationship is maintained.

2.3 Composites

Composites of the Equatorially averaged OLR were calculated based on propagating OLR minima (enhanced convection) every %°: of longitude accross the Indian Ocean and West Pacific (see Woolnough et al. (2000a) for the details). Figure 4 shows the time longitude composites for enhanced convection at 82.5°E (close to the centre of the Indian Ocean).

Figure 4: Time-longitude composite based on 36 propagating OLR minima at 82.5°E. a) OLR (contour interval 5Wm-2), b) SW flux (contour interval 5Wm-2), c) zonal wind stress (contour interval 0.005Nm-2), d) latent heat flux (contour interval 5Wm-2), e) SW+LH flux (contour interval 5Wm-2) f) SST (contour interval 0.05°C). In each plot the blue shading and dashed contours indicate negative values and the zero contour is suppressed.
The composites confirm the relationship temporal relationship between the convection, surface fluxes and SST, but also reveal the spatial structure of the flux anomalies. Associated with the enhanced convection in the Indian Ocean there is a reduction in the SWF, to the west of the convection there are westerly wind stress anomalies and associated with these there is enhanced evaportation and to the east of the convection there are westerly wind stress anomalies and in the west Pacific there is an reduction in the evaporation. Out into the central Pacific where the climatological winds are easterly the easterly wind stress anomalies are associated with enhanced evaporation. The surface flux anomalies lead to warm SST anomalies in the west Pacific, and following the convection cold SST anomalies inA the Indian Ocean. The composites show eastward progression of the convective signal and the flux anomalies and show a coherent link between the Indian Ocean and the west Pacific.

For convection in the Indian Ocean the SW flux anomalies contribute the largest component of the surface fluxes, but because of the phase difference between the SW fluxes and LH fluxes both contribute significantly to the surface fluxes. Similar componsites for enhanced convection in the west Pacific indiacte that although the basic structure of the anomalies is the same there are some differnces in the nature of the anomalies. For convection in the west Pacific there is little evidence of the easterly anomalies to the east of the convection nad hence there is no signal there in the latent heat fluxes, secondly the evaporative cooling has a comparable magnitude to the SW flux although beacuse of the phase differences between these two components there is little change in the magnitude of the surface forcing.

The spatial scale of the SST anomalies is such that there are about 60° of longitude between the maximum and mininmum in SST anomaly.Although the magnitude of the composite anomalies is small individual events can have SST anomalies up to 0.5-1°C. The magnitude of the composite flux and SST anomalies are consistent with the simple mixed layer depth caluculations with depths between approximately 10m and 50m, these depths are comparable with the range observed in the warm pool region.

2.4 Summary

The analysis of the observations has shown that the convection associated with the Madden-Julian Oscillation can generate SST anomalies on intraseasonal timescales, through the modulation of the shortwave and latent heat fluxes at the surface. Positve SST anomalies are found before and to the east, of the convection and negative SST anomalies following and to the west of the convection.

3. The organisation of convection by SST anomalies on intraseasonal timescales

(Woolnough et al. 2000b)

3.1 Experimental Design

To investigate whether intraseasonal SST anomalies, such as those observed in the Indian Ocean and west Pacific associated with the MJO, are simply an oceanic response to the convection or are able to organise the convection on intraseasonal timescales, and hence play a part in a coupled ocean-atmosphere phenomeon, a series of sensitivity experiments with an aquaplanet version of the UKMO Unified Model have been performed.

The aquaplanet configuration of the model retains all the physical process of the full GCM but the surface is ocean covered. A control integration is performed with a zonally symmetric SST profile prescribed, desgined to mimic the SST profile in the warm pool region. Perturbation experiments are then performed with prescribed SST anomalies with a magnitude of 1°C and a spatial structure (Figure 5) similar to that of the observed SST anomalies associated with the MJO. Experiments are performed with a stationary SST anomaly, and with the SST anomalies propagating eastwards with phase speeds corresponding to periods of 30, 60 and 90 days.

Figure 5: Spatial Structure of the imposed SST anomaly
For analysis the fields are composited relative to the centre of the SST anomaly, for the moving SST anomalies longitudinal diplacements can be interpreted as time lags.

3.2 The precipitation response

Figure 6 shows the equatorial precipitation anomaly from the zonal mean for the four experiments. For the stationary SST anomaly the maximum in precipitation is colocated with the maximum in SST, with suppressed precipitaton over the cold SST anomaly. For the moving SST anomalies the precipitation is shifted to the west of the maximum in SST anomaly. For the moving SST anomalies the faster the propagation speed the weaker the precipitation response.

Figure 6: Precipitation anomaly (mmday-1) from the zonal mean, averaged between 2.5°N and 2.5°S composited relative to the centre of the SST anomaly. Stationary SST anomaly (black line), and propagating anomalies with phase speeds corresponding to periods of 90 days (red line), 60 days (green line) adn 30 days (blue line). The thick black line shows the SST anomaly.
Examination of the Convective Available Potential Energy (CAPE) and Convective Inhibition (CIN) shows that the precipitation anomalies are not just a simple manifestation of the low-level instability. For the propagating SST anomalies the maxima in CAPE and minima in CIN (the most convectively unstable situations) are colocated with the SST maxima, indicating the low-level buoyancy is responds very quickly to the SST anomaly.

The westard shift of the precipitation and the dependence of the strength of the anomaly on the precipitation speed can be explained by considering the specific humidity (Fig. 7).

Figure 7: Specific humidity anomaly (g kg-1) from the zonal mean for each of the SST anomaly experiments, composited relative to the centre of the SST anomaly.
The boundary layer adjusts very quickly to the surface anomaly and the enhanced low-level instability leads to enhanced convection. However, entrainment of relatively dry air from the environment reduces the parcel buoyancy and inhibits the convection. As the convection moistens the environment, the effect of the entrainment is reduced, and more of the moisture carried in the convective plume is converted to precipitation. The time taken for the convection to moisten the environment explains the westward shift of the precipitation for the propagating anomalies. For the faster SST anomalies the convection has less time to moisten the lower troposhpere before the surface anomaly passes and the low-level instability is removed, leading to the weaker precipitation response.

3.3 The dynamical response

For the 30 day SST anomaly the dynamical response shows the characteristic structure of the linear response to tropical heating, with upper level anticylcones to the west of the convection and cyclones to the east. The zonal wind anomalies (Fig. 8) show the low level easterlies and upper level westerlies to the east of the convection, associated with the Kelvin Wave response. For the slower moving SST anomalies the structure is very different, the upper level cyclone pair is weaker and the anticyclones are shifted to the east. In the zonal winds, the Kelvin Wave signal to the east is much weaker, despite the stronger heating and the surface westerlies to the west of the convection are much deeper. At 30 days the forcing period is close to the natural frequency of the unforced moist Kelvin wave in the model and the forcing generates a large response in the Kelvin Wave, at slower freqencies the response in the Kelvin Wave is much weaker. The abscence of any resonance of the precipitation with the eastward propagating natural modes of the system suggests that the dynamical structure is a response to the heating associated with the convection rather than being responsible for organising the convection.

Figure 7: Zonal wind anomaly (ms-1) from the zonal mean for the 30 day and 60 day SST anomaly experiments, composited relative to the centre of the SST anomaly.

3.4 Summary

The model experiments have demonstrated that eastward propagating intraseasonal SST anomalies can organise convection with structures similar to that observed in the MJO. Furthermore the longer the period of the SST anomalies the greater the magnitude of the precipitation anomaly.

4. Implications for the coupled MJO

  1. The observations have shown that the convection can generate SST anomalies.
  2. The model integrations have shown that intraseasonal SST anomalies with a structure similar to the obsevered SST anomalies can organise convection on intraseasonal timescales
Both components of the system which are required for a coupled mechanism to work have been demonstared here. The timescale of such a coupled mechanism will be determined by the nature of the coupling.
  1. The strength of the precipitation increases for increasing period of the SST anomaly
  2. The strength of the surface fluxes will also increase for increasing period of the SST anomaly
  3. The timescale for the SST anomaly is determined by the strength of the surface flux anomalies.
  4. Long period SST anomalies will be associated with large surface flux anomalies and will tend to speed up
  5. Short period SST anomalies will be associated with weak surface flux anomalies and will tend to slow down.
The timescale of the coupled MJO will then have a prefered timescale which is determined by; the dependence of the strength of the convection and surface fluxes on the timescale of the SST anomalies, and the heat capacity of the mixed layer.

References

Emmanuel, K. A., 1987: An air-sea interaction model of intraseasonal oscillations in the Tropics. J. Atmos. Sci., 44, 2324-2340

Flatau, M., P. Flatau, P. Phoebus, and P. P. Niiler, 1997: The Feedback between Equatorial Convection and Local Radiative and Evaporative Processes: The Implications for Intraseasonal Osci llations. J. Atmos. Sci., 54, 2373-2386.

Hendon, H. H., and M. L. Salby, 1994: The life cycle of the Madden-Julian oscillation. J. Atmos. Sci.,51, 2225-2237.

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.

Madden, R. A., and P. R. Julian, 1971: Description 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.

Neelin, J. D., I. M. Held, and K. H. Cook, 1987: Evaporation-wind feedback and low-frequency varaiability in the tropical atmosphere. J. Atmos. Sci., 44, 2341-2348.

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.

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., and X. Xie, 1998: Coupled Modes of the Warm Pool Climate System Part 1: The Role of air-sea interaction in Maintaining Madden-Julian Oscillation. J. Climate, 11, 2116-2135.

Weller, R. A., and S. P. Anderson, 1996: Surface Meteorology and Air-Sea Fluxes in the Western Equatorial Pacific Warm {P}ool during the TOGA Coupled Ocean-Atmosphere Response Experiment. J. Climate, 9, 1959-1990.

Woolnough, S. J., J. M. Slingo, and B. J. Hoskins, 2000a: The relationship between convection and sea surface temperatures on intraseasonal timescales. J. Climate, 13, 2086-2104.

Woolnough, S. J., J. M. Slingo, and B. J. Hoskins, 2000b: The organisation of tropical convection by intraseasonal sea surface temperature anomalies. To be submitted.