Contents

• Section 1: Basic Structure and Typical Time Scale
  • 1.1  Overview
  • 1.2  MJO as Variation in Winds
  • 1.3  MJO as Variation in Surface Pressure
  • 1.4  Cross-Sectional Depiction
  • 1.5  MJO as Outgoing Longwave Radiation (OLR) Anomalies
  • 1.6  Intraseasonal Variations
  • 1.7  Upper Troposphere Velocity Potential
  • 1.8  Correlating OLR to Velocity Potential
  • 1.9  Typical Time Scales
• Section 2: Relevant Modeling
  • 2.1  Two-Layer Model of Heating at the Equator
  • 2.2  Schematic Cross-Section and Plan View
  • 2.3  Composite Stream Function Model Results
  • 2.4  Schematic Summary of Data
• Section 3: Tropical Weather
  • 3.1  Precipitation and the MJO
  • 3.2  OLR Anomalies and the MJO
  • 3.3  Tropical Storms and the MJO
• Section 4: More Effects
  • 4.1  Effects on Wind Stress
  • 4.2  Effects on Angular Momentum
  • 4.3  Effects on Jet Stream and West Coast Precipitation
• Section 5: Summary
  • 5.1  Presentation Summary
  • 5.2  Comment by Klaus Weickmann
• Bibliography
• Contributors

Section 1: Basic Structure and Typical Time Scale

1.1  Overview

Another thing I hope that you get form my description of the MJO in the tropics is its huge scale, it's a very large spatial scale disturbance; it’s also got very big amplitude, maybe the biggest variation next to the seasonal cycle in the tropics. So, even though after thirty years we still don’t know exactly how it affects the mid-latitudes, it’s so big in the tropics that it’s got to affect the mid-latitudes. There is a lot of work going on and I think it’s going to help for forecasts in the future.

Another important aspect is its time scale, and it’s got a sort of an average period on the order of 45 days, which means that it may help on say two or three week forecasts, and maybe even slightly more than that. Again, it’s probably a small percent of the variance, but still it's something in addition to what we otherwise might have.

So I’m going to spend most of my time talking about the basic structure, a little bit about the typical time scale, and I’ll mention something about some relevant modeling. It turns out that the MJO is at least consistent with some both simple and also general circulation models which put heating near the equator. It’s consistent with response to heating at the equator. I’ll show a little bit to give you an idea about some tropical weather and then some more sort of large scale effects.

1.2  MJO as Variation in Winds

First of all, the basic structure and the typical time scale. We're going to go back about 40 years for this data, this is winds at 150 hPa from Nairobi. Nairobi is about 35°E, on the east coast of Africa, it’s close to the equator, 1.3° South.

These are the winds in meters per second and it’s for two years, and we'll look first at 1964. We should say that a seasonal variation has been taken out, but there’s no other filtering, this is one rawinsonde observation per day, except in places where there’s missing data.

The vertical dashed lines are each about 50 days apart, to give you a feeling for the time scale, this is 365 days for 1964. You can see these fairly, actually very large variations in the wind, I mean you can see them just with the eye, you don’t have to do any kind of fancy statistics to bring them out.

1.3  MJO as Variation in Surface Pressure

Now we're going to go to the date line near the equator and these are pressures, they are taken actually from the reanalysis data, but we could have used a station data and shown similar results, these are in Pascals, so this appears like 4 hPa or 4

1.4  Cross-Sectional Depiction

Now, taking these various evidences together, we came up with this sort of cartoon to summarize what’s going on. Here again is the equator, here is where Africa intersects the equator, Indonesia, and South America, so this 360° of longitude across here, and each panel is different time, time is increasing downwards, and for say a roughly 40 or 48 day period the panels would be about 5 or 6 days apart.

If you look simultaneously at data in the mid-Pacific you get a picture of two circulation cells, which are converging in the lower levels and diverging aloft that support this convection. It moves eastward at something like 5 to 10 meters per second, and when the clouds get somewhere in the central Pacific (panel H), where the water near the equator gets cold, they seem to dissipate, and there’s some continued evidence that there are some clouds over South America later. But is not nearly as strong as what goes on the Indian Ocean and Pacific. So this is sort of the schematic picture or typical picture of what goes on in the equatorial plane.

1.5  MJO as Outgoing Longwave Radiation (OLR) Anomalies

This is the outgoing longwave radiation, and these are anomalies, shown in Watts per meter squared (W/m²). The blues are less W/m², which means the radiation is coming from a colder environment than say the reds or the yellows, so these are typically caused by high clouds and in the equatorial region it’s typically caused by convection, and maybe in the case of some of these warm anomalies we might be looking down at the ocean surface.

1.6  Intraseasonal Variations

Now, if we look with higher spatial resolution at these clouds, we can see that these structures are very complex. But Nakazawa wrote in the journal of the Meteorological Society of Japan back in 1988, he looked at very high resolution satellite data and he called what we just saw in the last picture, he called those intraseasonal variations, and they are sort of indicated by this outline here, and then if you look with high spatial resolution (magnified view of the area in the yellow rectangle), inside this intraseasonal variations you see more narrow bands that are moving eastward.

And this is on a time scale of 40 to 50 days, and here is day 20 or so and we come down about 40 days later, day 60, so these intraseasonal variations have this 40 to 50 day time scale, actually is broader than that, and we’ll see that in just a minute.

He then called these structures inside the intraseasonal variation Super Clusters, and then if you go down to even higher resolution and you look at a Super Cluster moving eastward, like the magnified view of one of the heavy black lines, we see inside the Super Cluster convection which actually moves westward, and it has a time scale of about a day. For some reason you have this large convection and then you have developing convection always to the east of the existing large convection, so that gives the eastward movement of what Nakazawa called the supercloud clusters. So it's quite complex cloud structure involved in these things.

If we go back to the cartoon again, we see that in the upper atmosphere we have this divergence above the convection, and so we might expect divergence above these low OLR values.

1.7  Upper Troposphere Velocity Potential

One way to look at the divergence is to look at the velocity potential. This now is the same time period as the Outgoing Longwave Radiation, but it’s the velocity potential and again here is the equator from Greenwich to the date line to Greenwich so this is eastward going toward the right, and the time goes from mid-September of 2001 up to mid-March of 2002, so time goes downward. These orange areas are high values of velocity potential, and the blues are low values, and the diverging wind is proportional to the gradient of this field, so that the diverging wind is blowing from blues to the kind of yellows or reds. So you expect maximum divergence somewhere in these blue areas, and again this is the upper tropospheric velocity potential. So now we can compare this. We are looking at this is because the two variables that are used a lot to identify these MJOs are the outgoing longwave radiation and/or the velocity potential, and now we can compare them.

1.8  Correlating OLR to Velocity Potential

You can take the total winds and brake it up into a stream function and a velocity potential, and the winds associated with the stream function have no divergence. They're strictly kind of a circulation and the component that’s associated with the velocity potential is all divergent. So it’s kind of a mathematical breakdown which gives you an idea of where the maximum divergence is.

On the left hand side we have the velocity potential and on the right hand side the OLR anomalies.

We can pick, for example here is one of these eastward propagating MJOs, and maybe close to the date line, at about 16 December, mid-December, there’s lots of low values of OLR, which presumably comes from tall clouds, which in the equatorial region is mostly convection. And if we go to mid-December on the velocity potential we have an area of blues.

These are negative values of the velocity potential and again the divergent wind is proportional to the gradient of this field, so that the winds are blowing from these negatives, from the blues out to the yellows, so there is divergence.

We can see another MJO between January and march (red lines), you can see eastern propagation, and the first one, between September and October, (blue lines) has some corresponding divergence aloft. When people talk about these MJOs, when they identify them, it’s typically either looking at OLR that’s moving eastward or this velocity potential, that tends to move eastward, with this roughly 50 day time scale.

1.9  Typical Time Scales

In fact the next picture gives us an idea of what the typical time scales are. I’m going to go back to, I think it’s the first slide just to show… OK, this happens to be Nairobi, and we’ve got these big oscillations in the wind and the histogram that we are going to discuss next, the way that I got that was. actually I didn’t use Nairobi I used Truk Island, which is at a 152°E, so it’s closer to the date line, it’s in the western Pacific, it’s about 7°N and it has similar variations in the zonal wind. And I just subjectively estimated the time between, I can’t remember it was either successive minima or successive maxima, to give some idea of the average time scale of the variations.

Here we have days between successive minima of the Truk 150 mb zonal wind, and this is the number of cases and I can’t remember exactly how many years of data this involves, but here is March, April, May, June, July, August, September, October, November, and December, January, February. The mean is 45, 45 and 48 days in September, October, November and 45 days in December, January, February so it’s near 45 days but you can see that there is this very wide spread; there is one that is in the order of 22 days, or something. This is one case, and there is another one that is in June, July, August that’s of the order of 80 days, almost as long as the season. Now this is pretty subjective, but it gives some idea of the spread of these phenomena; in fact, in his work Klaus called it 30-60 days, which I think is a pretty good time scale, it pretty well takes into account the wide range of periods that we see with these oscillations.

As we mentioned, they’re not always there, but it turns out that this number of cases is about 60 % of the total cases that could have been typically each season, so that it’s there about 60 % of the time. Although you can use some other variable to estimate it, and you get bigger numbers, upwards of about 75 % of the time, so the oscillation's there somewhere in the neighborhood of 60 to 75 % of the time. So that’s sort of the basic structure and the typical time scales.

Back to Top

Section 2: Relevant Modeling

2.1  Two-Layer Model of Heating at the Equator

We will now look at a couple of relevant modeling experiments, and one is a simple experiment by Adrian Gill that was in the Quarterly Journal, where he had a two-layer model, a low level and an upper level. And he put heating at the equator, so the yellow area is supposed to indicate where the heating was, and then he looked at the response.

So this heating could be like that convection we see in the MJO, and he looked at the response to this heating and what happens is you get a response to the east of the heating that’s all in the zonal wind, in the east-west wind, and this is like a Kelvin wave, which was mentioned before. And Kelvin wave is after Lord Kelvin, but it’s also called an "edge wave," in the ocean for example when you have a coast, it’s a wave that moves parallel to the coast and it needs an edge or a wall. It turns out in the atmosphere that the equator, because of the changing Coriolis parameter across the sign of it, across the equator acts as a wall, so it’s sort like of an edge wave in the ocean.

It’s entirely in the zonal winds and you don’t see any meridional component in the Kelvin wave. In low levels you have convergence into the heating, and in upper levels you have divergence out of the heating, on the east side.

On the west side you get two Rossby waves. The area inside of the red line represents a low pressure anomaly, and in the upper troposphere inside the blue line we have a high pressure anomaly.

The red arrows mark circulations like this, which is a Rossby wave to the west of the heating, and there’s a symmetric Rossby wave on the other side of the equator, so in this case where the heating is right on the equator you get these 2 symmetric Rossby waves forced to the west of the heating and a Kelvin wave to the east.

It may not happen everywhere, but you can see that the winds are approximately geostrophic here. These winds are coming from the east into the low pressure anomaly, and so the high pressure in the northern hemisphere is to the right, the winds are about in geostrophic balance; and similarly, in the upper levels the winds are diverging out of this high pressure area. So in the northern hemisphere, anyway, the highest pressure is always to the right of the winds; they’re about in geostrophic balance.

In the section we'll see a kind of a cartoon which sort of summarizes some evidence that we won't present fully, but which indicates that at least when you look in small scale near the equator the MJO looks something like this simple model of Gill, a response to heating at the equator.

2.2  Schematic Cross-Section and Plan View

This is from the sketch we saw earlier and it just shows when the convection say is around Indonesia. Here is a low pressure anomaly that's reached out to the east, and our circulation cells. And then up here is in plan view, so it’s the upper troposphere looking down on this cloud, and indeed there is a response or at least data indicates that there is this zonal wind diverging out of the heating, which in this case would be based on the outgoing longwave radiation, with suggests that there's been some convection.

Most of the time the heating is not right on the equator, it’s in the ITCZ. So I’ve drawn this with the heating, this little plus is slightly into the Southern Hemisphere and so you don’t get the 2 symmetric Rossby waves, but you get an asymmetric Rossby wave. And you have these surges of winds blowing across the equator, as shown by the blue arrows.

This shows a certain degree of consistency with Gill's simple model.

2.3  Composite Stream Function Model Results

2.4  Schematic Summary of Data

Here is a schematic from Klaus Weickmann, 1983. Again this is based on data that he looked at and we won’t show fully, but we will provide a summary.

Here is the cloud that’s associated with MJO and in large scale you’ve got these anticyclonic eddies (blue arrows), Rossby waves sort of even with and to the west of the clouds, and out to the east you get flow in this sort of like around lows. It’s sort of like a Kelvin wave near the equator, it’s all in the zonal wind, and then there is this return flow at higher latitudes (green arrows).To the north, the polar jet, has expanded down into lower latitudes in this region, with the winds flowing from the west, and then the polar jet tends to be contracted in this region (red arrow) , out to the east of the clouds, so that there are these very, very large scale responses to the clouds in the tropics.

If the picture was always like this we could probably really use this information right now for forecasting over United States, but in some sense maybe we could think of this as an average picture just like the averaged period which we saw in that histogram, there is an averaged period but there’s a wide spread.

So Vern used the word 'flavors', so there’s lots of flavors to this thing, and we’ve got to sort of sort through those before it's going to be useful for forecasting. OK, so that’s a couple of experiments and to show that in fact that the MJO seems to be fairly consistent with what we know is the response to heating near the equator and the heating is this large convection, presumably..

Back to Top

Section 3: Tropical Weather

3.1  Precipitation and the MJO

Let's take a look at tropical weather. Here we are back again to the summer of 1979, as with the Section 1.3 graph showing the large variations in the pressure, which were at the date line. This now is over the Arabian Sea. The line is the precipitable water measured by a satellite over the Arabian Sea, which is to the west of the Indian subcontinent, and the dark bars under the line show the rainfall over a number of stations up the west coast of India.

The rainfall here is in cm per day, and you can see that here is May, June, July, August, and September and you can see maybe this is before the onset of the monsoon. Somewhere in here the monsoon gets going, and you have this very large precipitation, and also an increase in precipitable water over the Arabian Sea, and the precipitable water is from the surface up to about 700 mb. And then the precipitation stops. This is sometimes called the break in the monsoon. It doesn’t quite stop, but it reduces quite a bit, and then it goes back up again; this is an active period in the monsoon. And then it goes back down again. And so one maximum is around the 20th of June and the second one is about the first of August so that’s about 40 days.

3.2  OLR Anomalies and the MJO

If you look only at the right hand side of the image for the time being, this is sort of like that OLR picture that we looked at earlier, I mean it is the OLR, except that this is from a paper from Lau and Chan. Along the bottom you can see the Greenwich meridian, the date line, and 120W, so it doesn’t go all around the world. The time spans from May through October of 1979.

These are OLR anomalies and the dashed contours represents low OLRs, so that’s presumably convection, and you can see here is probably an MJO that moves eastward, and the line at 80°E shows about the longitude where that precipitation was measured. This is along the equator, within 5º of the equator, and we can see three probable MJOs (green lines). So you can see the eastern propagation just like we saw in those other OLR pictures.

Now, on the left hand side is not a cut from east to west but from 50ºS to 50ºN and it's along this 80ºE. So what Lau and Chan did was look at the clouds then to see if there is any kind of propagation north or south. This dark area is the heavy rain in June, around the 20th of June, I think it was a maximum we saw in the last transparency, and here is the heavy rain in mid-July to mid-August. So these are 2 rainfall periods, and those stations are along India, so they’re probably from 15 to about 30ºN or 15º to 28ºN. So when there's this precipitation you can see at about say 20ºN, these look like northward propagating clouds, this negative anomaly, and it goes sort of with, at the same clouds are moving eastward.

So I picture it sort of as the wake of a boat. You have the clouds moving eastward, but also there is this northward propagation up across India and they cause the active periods, and then there’s a break somewhere in here, and then a second active period, and probably a third. As I was preparing this I was thinking I should have looked at precipitation, this only goes to the 15th of September. Although the monsoon may stop here, if the monsoon is still on, there should maybe be another maximum in precip sometime in late September with this cloud that's moving north and simultaneously as its westward movement. So that’s some of the weather right in the Indian Ocean.

3.3  Tropical Storms and the MJO

Here is where we come to evidence that the MJOs affect tropical storms. There are 7 panels, and they’re about typically 5 days apart. This is kind of a composite and the shading is the velocity potential and the green are negative velocity potential regions, so those in green are upper level divergent areas, so there’s upper level divergence. The equator goes across the center of each panel, and the panels go from 30ºN to 30ºS, and are about 5 days apart. It's kind of a composite of many MJOs and it shows this divergent area going eastward (red line). We've outlined South America and Australia in the first panel to show that the sequence actually starts with the divergence in the far western Pacific. The diagonal line sort of corresponds to the clouds, although the strong clouds typically only make it to about the date line, but then in the upper troposphere you sort of get a wave that continues all the way around the Earth.

The ovals in the image on the left are points of the origin of tropical systems that develop into hurricanes, so most of these little circles form under this divergent area (upper level divergence area) caused by the MJOs as they move eastward.. This got a lot of publicity I’d say last summer or so, maybe just before the hurricane season started, because here is the Atlantic, you can really notice it say in the western Pacific, and then also these hurricanes that form off the west coast of Central America (panel 4), but there’s some also in the Atlantic. Maloney and Hartmann's work, at least one of their papers concentrated on the Atlantic hurricanes and there seems to be quite a difference between numbers of tropical storms that form under this MJO divergent flow, versus the brown area which is upper level convergence, there's very few formations in that region, so this seems to be maybe an important effect..

Back to Top

Section 4: More Effects

4.1  Effects on Wind Stress

Finally, we're going to talk about a couple more effects. Again this is 1979 and this is the wind stress over the tropical Pacific. We show this because it gives an idea of what large scale variations occur due to this convection.

First of all, this is average of June, July and August data, and the heavy black line is the averaged wind stress. The wind stress is the wind blowing on the ocean, which puts a friction on the ocean and tries to drive the ocean currents in the direction of the wind. The wind stress is typically parameterized by the wind strength. Where the wind stress is positive (to the right of the zero line), it means the winds are from the east, by convention. Interestingly enough the winds are slowing down, they’re picking up momentum from the rotating Earth, so they’re slowing down the rotation of the Earth, and they’re increasing the angular momentum of the atmosphere. Easterly winds, by convention are positive, because easterly winds… the Earth is rotating against the easterly winds and imparting a westerly momentum to the winds, they’re given momentum from the west to the east, and you’ll see that in a minute, in that atmospheric angular momentum changes.

Remember on the 20th of June the convection (dotted line), the rainfall was very strong over India. Now this is the Pacific, so this is quite a ways away from its rainfall over India, and yet the winds are strong from the east, the trade winds are very strong. Remember, positive means strong trade winds, and this is just by convention, if we had used the average winds, of course it would have been negative, because the winds from the east are very strong.

And so you have this convection out near India and the entire Pacific from 30ºN to 30ºS has increased strength in the trade winds, and you get this wind stress which in fact is slowing down the Earth and they can measure changes on the rotation rate of the Earth. It’s slowing down the Earth, but it's picking up angular momentum to distribute into the atmosphere, so the angular momentum of the atmosphere is increasing, because the total angular momentum of the atmosphere plus the ocean plus the Earth, remains essentially constant on these time scales. But anyway the stress is much bigger than the average stress.

Now on the 10th of July (dashed line) that was a break period, that’s when it wasn’t raining, that was before the second rainfall that started sort of the end of July into August, and at that time the wind stress is less, at least at most latitudes, than the average, and where it goes negative that means that westerlies have dipped in, westerly winds at the surface now have replaced these very strong trade winds from the east. So there is this very big variation, and this is averaged over the entire Pacific.

4.2  Effects on Angular Momentum

Here is the atmospheric angular momentum, this is sort of a measure of the strength of the westerly winds averaged over the entire globe, and the dots are just sort of a smooth curve showing the seasonal variation. Here is May, June, July, August again of 1979, so there’s a minimum in the summer in the total atmospheric angular momentum, and so you can see this seasonal variation with the minimum. But then when you look at the real values, this was either 1 or 2 values per day, you see this variation; and again here is times of maximum precipitation over India, so that’s when the clouds were forming in the Indian Ocean of this MJO that moved eastward subsequently. And the angular momentum then has this same period. This is a maximum say sometime in late June, and a second maximum in say late August, so this is maybe 50 or 60 days between these 2 peaks, and maybe this is another maximum here that goes with that MJO. I mentioned that the total atmospheric angular momentum of the Earth atmosphere and ocean remains constant, so if all of that angular momentum goes into the Earth some change in the angular momentum of the atmosphere corresponds to a change in the length of day, and this little bar here represents a tenth of a millisecond (10-4 seconds) change in the length of day, which can be measured, geodesists can measure these things.

These variations, then, are on the order of a tenth of a millisecond, actually the amplitudes are maybe 2-3 tenths of a millisecond, and the angular momentum is increasing when that wind stress was very strong, when the precipitation is very big so the angular momentum increases and in both cases when the precipitation is big and these easterlies are very large, the Earth is rotating against the easterlies, so the easterlies are slowing down the rotation of the Earth, so they’re making the length of day slightly longer, on the order of a few tenths of a millisecond. The atmospheric angular momentum is increasing, so the angular momentum is coming out of the solid Earth and going into the atmosphere. We used this to show that this MJO shows up in globally averaged variables.

4.3  Effects on Jet Stream and West Coast Precipitation

One last effect, I take this also from the Climate Prediction Center webpage, is the effect occasionally on heavy west coast precipitation events.

The cloud represents the MJO, first over Indonesia, and then moving out to the east and then maybe over about the date line, in the east, in the third panel. So the heavy precipitation event occurs about at the time when the clouds are out in the central Pacific (panel 1), so 7 to 10 days before the event, when the clouds are further west, and this again sort of an average picture, it doesn’t occur all the time, but there can be some kind of a blocking high with a strong polar jet which arches over the high and doesn’t affect the West Coast of the United States.

Then as the clouds move a little further east (panel 2), the jet starts to split and you get a trough here replacing this high, and then eventually the trough dominates the West Coast (panel 3); and the strongest jet is along way further south from this one here, and you get the precipitation on the West Coast.

I think it was George yesterday showed some satellite pictures of clouds sometimes called the "pineapple express" because they tend to come over Hawaii, so maybe this is like a pineapple express that’s affected by MJO.

Back to Top

Section 5: Summary

5.1  Presentation Summary

To summarize, there are these large and they’re very complex cloud systems, and they move say from the Indian Ocean to the Central Pacific, and they move at time scales of the order of 5 to 10 m/s.

The local time scales range, you saw that in the histogram there is a wide range, but the average is like 45 days; typically the time scales fall within a 30 to 60 day range.

The basic properties of the MJO are consistent with the response to heating near the equator, at least what we know of it based on both a simple two-layer model, and also the more complicated general circulation models, and we looked at an example of the Indian rainfall and also tropical storm formation that are affected by these MJOs.

And then the globally averaged quantities like the atmospheric angular momentum, and then also the mid-latitude weather, which is certainly affected, because this thing is so huge, but we've got to sort out exactly how various kinds of MJOs affect mid-latitude weather, if we’re going to use for forecasting.

5.2  Comment by Klaus Weickmann:

"The MJO appears to have active and inactive periods. For example, the last 20 years appear to have been more active than the previous 20 years. In other words, the MJO seems to have a high degree of variability in terms of its magnitude."

Response to comment:

One measure of this phenomenon is the spectrum, which you heard about yesterday too, which is a measure of the variance as a function of frequency, and of course the spectrum for this phenomenon typically it's sort of smooth, red noise, that is more variance at low frequency, less at high frequency, and smoothly varying. And there's usually sort of a peak, some extra variance in this 30 to 60 day range. But along with what Klaus said, if you look at different periods, I can't remember what period we looked at, the peak definitely was around 30 days, where typically it maximizes around 45 days. The period of it changes as well. And I forgot to mention the seasonal variation, it's more strong in the winter and early spring than in the summer, although you see the variations, in the summer as well, but on average it's stronger in the winter.

Back to Top

Bibliography

Back to Top

Contributors

Back to Top

Congratulations!

You've completed the Madden-Julian Oscillation Life Cycle module. Now you can take the Module Quiz to see how much you've learned. And don't forget to fill out our User Survey to send us your comments on this module.