 And mainly, I will show results from a dry model experiment that shows the two-way interactions between MGO and the North Hemisphere, because there is no tropical processes, no connections, and this one yesterday Christina mentioned, but today I'm going to show that in more detail. And after that, I will talk about the two-way interactions between MGO and NO, and we'll again focus on the NO impact on MGO. So I should mention that the extra tropical influence on MGO is less well understood compared to tropical influence on the exotropics. There's a lot of processes, mechanisms that are not well understood, so it's an ongoing topic. So for the examples of mid-latitude influence on the MGO, there is one started by Ray and Zhang in 2009. What they did, they used a tropical tunnel, a channel model, so only in the tropics. They look at two MGO events to see what is the initialization process of those two MGO processes, events. So what we found is that the only factor that is critical to the production of the MGO initiation is the time-varying and matter-boundary condition that is in the mid-latitude. So they specify the boundary condition from reanalysis, but when such a matter-boundary condition are replaced by time-independent conditions, for example, climatology, they kill the MGO, so there's no generation of MGO. So these results just indicate that the tropical influence can be a very efficient mechanism for the generation of MGO, so that is one of the example early study. And later, they look at the mechanisms, what is the processes, they look at the momentum transport and they found that the momentum transport from the mid-latitude has a big contribution to the MGO initialization. Another example is by Hong at all, that's a very recent paper, they look at the exotropical forcing for 2015 MGO and El Nino events, and they found the southward penetration of the northward wind that is from the mid-latitude propagated to the tropics that influence the generation of MGO and El Nino in this period. And then later, Nick Hall, he will talk about more examples and his study on the mid-latitude impact on the MGO. So now I will switch to talking a little bit more detail about the tropical, exotropical interactions in dry GCN. So the model we use is a primitive equation atmospheric GCN, not to was developed by Nick Hall, I'm glad he is here this time. I think he will describe in more detail this model tomorrow. And this model is T31 and the 10 levels model is quite low resolution but able to capture the large-scale phenomenon in the atmosphere. And the special in this model is it has a time independent forcing that is calculated empirically from the real analysis, long time day-to-day analysis. So this time independent forcing is to maintain in the model a winter climate college. So that means the model, if you run a long time, you do the climate average, you get a very similar climate college to the observation. And also that's only the time mean but also the challenge, the variabilities. So because it's time independent forcing, no other variation in the forcing. So all the variabilities generated in this model come from the internal dynamics of the model itself. So what we see is mainly the, for example, the barricannitability, the methanatute, that will, it's main part of the variability in the model. There is no moisture and no interactive connection. So it's a dry model. So what we did, we run the model for 3600 days, it's because of the forcing, it comes from the winter condition. So the integration is a perpetual winter integration. So we don't use the first six days that is ignored as the spin up. So we use 3600 days to do the analysis. So what we would expect is that there won't be any variability or waves in the tropics because there's no like moisture or little connections. There's no, but what we found is quite interesting is like this. So this is a model result for the 10 degree south to 10 degree north average of 250 millibar velocity potential. So that represented the divergence field in the tropics. So this is the diagram, this is time from zero to 200 days. This is a longitude. So what you can see is that in this period, you see the eastward propagation of the disturbances for velocity potential. So the propagation is eastward. And this is another period from, OK, so that's working. It's OK, just if you have that. Thank you. So this is another period from day 2800. Again, you see this kind of eastward propagation. So this is the raw data, this unfiltered data. If you do a 20-day to 100-day band pass filtering, you plot this diagram again, you see this kind of nice eastward propagation of waves. One feature is that the wave is stronger in the eastern hemisphere. And also it propagates slower in the eastern hemisphere than in the western hemisphere. So those features, it looks very much like the MGO. So when we got these results, we are intrigued by these results. Why did this happen? So as I said, this kind of model, the only thing in the atmosphere in the model is the intero dynamics. That is those kind of waves generated is the only reason that's from the middle altitude. So the time interval, if you do the two of this, if you do the wave number frequency spectral analysis, you see that this is the velocity potential at 250 millibar in the model. You see this peak is wave number one. And the frequency is about 30 days. So it's a little bit too fast compared to MGO. So you will see later it's Kelvin wave generated in the tropics. Compared to the observation, the velocity potential at 200 millibar, this is slower. It's about 50 days. That is really MGO. So that's the main structure, the wave number one. And it's close to this MGO. So what we did, we did the UFO analysis, tried to capture this wave. And for velocity potential, this is a UF1, UF2. You see the monopostructure and the dipostructure. And the PC2 lead PC1 by about seven, eight days. So that means this happened about seven, eight days later. This pattern will appear. That represents the eastward propagation. So we designed a very simple index. It's the PC2 plus PC1, eight days later. So it's like a combined consideration of these two indices. So we produce metamattitude and a tropical combined diagrams to visualize this process. So I will go through this quickly to see how the tropics is connected to the metamattitude. So the colored area, that is the velocity potential at 250 millibar. And the contour, that is 250 millibar stream function. So the stream function that will represent a Rosberg wave in the metamattitude. So if you look at the color area, you see this with time, you see the eastward propagation. But now we go back. If you focus on this metamattitude wave trends, you see at the beginning, there is a wave train in the North Pacific. But you will see with time, you will see this energy propagate to the Atlantic and then come back to the tropics. See, this is time. So now increase in the Atlantic. So there is some indication for the NNO generation. And now here increase the wave activity here. And also at the same time, the tropical divergence field intensifies this area and then propagates eastward. So this kind of like a two-way interaction, it seems it's quite clear from this kind of simulation. So we calculated the wave activity flux for this along the field of stream function using the formula by Pacaya and then Nakamura. So what you can see is at day minus 8, you see the wave activity flux from the North Pacific, from tropics into the North Pacific to North America. And eight days later, you see the wave activity, a large part of flux toward the tropics. So those kinds of wave activity will influence the tropics to generate another cycle of the tropical waves. You also see that another branch of wave activity that's eastward probably that will influence the minimum altitude Eurasia component. All right, so this is a demonstration of the dry model. So what we found is the tropical intracellular ability to generate it in the dry model. Tropical interaction are crucial in generating this model of our tropical variability. And that indicates the exotropical influence on the tropical waves. So the remaining question after this study is the contribution from moisture and connection. So probably I talked with Nick before. We can do another experiment with including moisture to see how this model behave after we include another microprocessors. And for the mechanism is how do exotropical large-scale disturbances that are equivalent to barotropic propagating the tropics to generate the tropical waves, not to baroclinic. So this is an area that's not very well understood. Right now, we are doing some kind of theoretical work to understand how this interaction happens. There's two different kinds of waves that can be influenced to each other. All right, so now I will switch to another topic. It's not a topic. It's related. It's the NGO and the NNO interactions. It's based on earlier study and also David mentioned also. So we use the NNO index, that is a pentad average. So that's a 5-day-5-day average. And the NGO index is the Weir and Hinden index. So we also have the pentad to pentad average. And the purpose is to look at the intracellular variation of NGO and NNO, how they are correlated and connected. So we look at the winter period. So because the interactions under the wave activity, roughly the generation is strong in the winter season. All right, so this shows the eight phases of NGO, phase one to eight, and the equatorial region from 15,000 to 15,000 north. So what we can see is phase one, you see the anomalous precipitation generated in the western Indian Ocean, and that's perforated eastward. So I want to draw your attention to the phase two, phase three, and the six, and the seven. That is, when there is a connection dipole structure, the enhanced precipitation in the Indian Ocean and the reduced precipitation in the western Pacific. It is in those phases that will influence the middle attitude the most. The phase two, three, and the six, and seven, they are just opposite. So this is the, if you look at the debug heating, this is cooling, this is the heating. So what we did, we look at the lag composites of NL index for different phases of NGO, one to eight. And the lag, zero, that's a simultaneous composite. For example, this is the composite when NGO is in phase four. And the positive lag means NL lags the NGO. The negative lags means NL lead the NGO phase. So if you look at this part from zero to positive, what you can see is when NGO is in phase two, three, four, that I just showed, that's a dipole in the tropics, you see this after one, two, three pentas, you see the probability of positive NL. So those numbers, they are 95% statistically significant. Those white areas is not statistically significant. I didn't plot. So the positive, the red indicates the positive NL, the blue indicates the negative NL. So that means when NGO happened in two, three, four pentas, so two, three pentas later, you see positive NL. And when in six, seven, eight, you will see about two to four pentas later negative NL. So it's very statistically significant results. And other than that, if you look at when NL leads the NGO, you will see that when negative NL happens about three to five pentas later, you see the occurrence of NGO two and three. And when positive NL happens about three or five days five pentas later, you see the probability of NGO phase six and seven. So this is like a two-way interactions. The NL and NGO, not only there's lag, and also there's lead association. So this one David showed already, you see the phase three, you see the zero pentas, one pentas, two pentas later, you see the development of a positive NL. And phase seven, that's the opposite. You see the negative phase of NL. So I just take a little bit of time to show you why this pentad and those dipole structure affected the middle attitude is the strongest. So what we did, we used the ORR, a pentad ORR to do UF analysis, that is to find the NGO, that is another representation. It's very similar to a wider index. But now we use only the ORR, because ORR is more connected to the convection. So the first UF is like a convection in the maritime continents area. But the second UF is a dipole. It's a convection in the Indian Ocean and a depressed convection in the West Pacific. So we'll see that which one is more effective in forcing the NL. So we designed thermal forcing, and the first experiment is put a heating in the maritime continents. And the second experiment is put a heating in this region and a cooling in this region is a dipole. It's trying to mimic this second UF. So this is the results for the model integration average for 500-meter-bar height anomaly, 6 to 10 days later. This is 11 to 15 days after integration. You see experiment one, the response is very weak. Almost there is no response. But experiment two, you see the response is very strong. And you do see there is indication of positive NL. So to understand this, we did a lot of experiment to put this heating at different locations to see which one is the most sensitive location for this response. So again, we do a linear integration for winter-based state with a single center heating source put on the equator for this region, 60 degrees east to south, 150 west, a 10-degree interval. So we have 16 experiments. And we look at the response at day 10 to see the sensitivity to the location. So here is the example. You put the heating at eight east, so that means the Indian Ocean, tropical Indian Ocean. You see positive anomaly response in North Pacific. So this is similar to the negative NNO response. So the response is not very sensitive to the location of the heating in this region from 60 to 100 in that area. So that means like the question David raised in the last lecture is the heating is moving. How can you follow this kind of response? But it's not really sensitive to the location. So in this region, the response is very much like this pattern. When you put the heating at 110 degree, that's nearly a maritime continent. There is almost no response. It's very weak. But when you put the heating at east of this area, the response is very similar to this pattern. So it's not very sensitive. But if you compare this pattern with this pattern, you can see they are almost opposite. So the response, if you put the heating in this region and this region is opposite. But as you see for the MGOFIS-3 and the FIS-7, they are dipole. There is a heating and a cooling. So that means because this experiment is all heating there. But if you switch this to cooling, it's the same as this one. So that's why you add this two pattern together. You got a very strong response. So that's why the response to MGOFIS-3 and FIS-7 is the strongest. So that's why the dipole structure. Now also Andy showed Friedrich's study to look at different centers forecast with respect to MGOFIS-3. So the reason for this is that the heating location with respect to the middle altitude jet. This region is, for example, in the middle altitude, the East Asia jet, the core is just close to Japan and East Asia. So this region is to the west of the jet. And this region is to the east of the jet. So this kind of behavior is really relative location to the jet. All right. So again, we can calculate the wave activity flux to see the MGOFIS-3 simultaneous wave activity flux. You see the northward propagation of waves. And one pentat later, you see more to the north east. And two pentats later, you can see the wave activity flux is many in the Atlantic. And there is a strong branch to the south. So this kind of wave activity to the south that will impact the tropics probably will generate another cycle of the MGOFIS-3. So what we did is also to look at the MAC regulation to NNO to see how the NNO impact the tropics to see the zonal wind how to propagate southward. So this is when the NNO is at the strongest, that's the MAC-0. And one pentat later, you see this kind of amplification of the zonal wind in the sub-tropical Atlantic region. And two pentat later, you see stronger that there is negative anomaly of the zonal wind in the equatorial tropical Atlantic. And again, you see this kind of movement. So that's what the last diagram shows the lag propagation of the zonal wind. If you compare to the composites of the zonal wind for different MGOFIS, you can see that when the NNO happens about four, five pentat later, it's very similar to the phase seven and the phase six and seven. So that's why in the table, I just show you at the beginning, the MGO lags the NNO about four to five pentat. So this kind of southward penetration of the zonal wind. So schematically, it's the MGO propagates eastward, send the Rosberg wave to the North Atlantic. And probably there is interactions with transients in intensified NNO. And then there is an influence to the topics. So now I'll show you some examples how the interactions between MGO and NNO impact the substance in the prediction. So this is one experiment we did to look at the NNO forecast scale for weak MGO and the strong MGO comparison. So the red curve, that's for when the initial condition has a strong MGO, you forecast the NNO. And the blue is when the initial condition has a weak MGO. So what we can see is that there is clear separation. The strong MGO initial condition produce a better forecast scale of NNO. So this indicates the tropical influence. And also, you can separate this forecast by the phase of MGO because what we saw is that the dipole phases produce the strongest impact. So we category the forecast with MGO phase 8, 1, 4, 5. So that's not a dipole. Compared with forecast with initial condition it has a phase 2, 3, 6, 7. That has a dipole. So you see the NNO scale, it's better when you have a dipole intercondition. So it's very consistent with the analysis. So this just to show the scale of 500 meter bar height for Pentas 3 and 4. This is weak MGO and the strong MGO. You see the increase of scale in the North Atlantic region. And also for temperature, you see this is a weak MGO, strong MGO. You see the better scale in this region and the European area. And this is the difference. So strong MGO minus weak MGO. So this diagram from Friedrich Wittar he looked at the different models. This is another demonstration Andy showed the diagram this morning. This shows the green one shows when the MGO is in phase 3 after three Pentas, the NNO index. So this line is the real analysis. So that's the expected observation. So most of the model has a weaker response of NNO after MGO phase 3. On this side, it's the three Pentas later after phase 7. Again, you see most of the model is weaker response. But the Environment Canada model is very close to the observation. So again, we look at the S2S handcast data to compare the NNO scale for dipole and the lung dipole. This is the dashed line that's phase 2, 3, 6, 7. That's the dipole MGO initial condition. And the solid line, that's the lung dipole. So you see the difference. It's quite clear the MGO phase when there is a dipole, you get a better forecast. So now we can look at the other direction, the impact of the NNO on the MGO forecast scale. So we separate this forecast, look at the initial condition when there is a strong NNO. And here is the scale of MGO. So what you can see, this is our model forecast compared to the persistence. The MGO, there is a scale compared to persistence. But this one is the initial condition has a weak MGO compared to a strong MGO forecast. So the bar, so we have a lot of ensemble members. So there is some period where we'll separate it. You can see in the first 10 days, there is almost no difference. But after 10 days, you see strong NNO forecast, the scale of MGO is better than a weak NNO. So because you see the lag at this table, it's about 3 to 5 pentatons. So that means the impact is up here after 10 days. So this really indicates that the mid-natitude signal has an impact on the MGO forecast scale. So there is two-way interaction. So now we look at the scale distribution for 200 millibar zonal wind scale near the tropics for strong NNO. You see that this scale, there is quite a lot of scale in this band, the tropics. But when you initialize the forecast with weak NNO, the tropical zonal wind scale, there is a lot of places is not as good as this one. This is the difference. So this is the increase of scale if you have a strong NNO in the tropics. OK, this one is the average of scale between 0 and 90. So that's mainly just the south of the North Zatomantic oscillation and the impact in the tropics. So this is the zonal wind. I think this is the near equator. This is as a function of latitude. So this is the time, lead time. What you can see is in short-range forecasts, the middle latitude, the scale is better in the tropics because we have the theory of barricade instability to support this short-range forecast. In the tropics, there is a lot of convection, precipitation, so the scale is very low. But if you go beyond 10 days, you see the scale in the tropics is better than the middle latitude. So this is the average scale. But you can categorize the strong NNO forecast and the weak NNO forecast. This is the difference. So what you can see is the strong NNO forecast. You have a better scale than weak NNO forecast after about 10 days. It's mainly in the tropics. So that means the NNO, strong NNO, really send a signal to the tropics, impact the tropical prediction. So this is why this happened in the observation. You have a strong NNO, this is composite, lack of composite, 200 millibar zonal wind alarmony after a strong NNO. So the NNO, you have a strong zonal wind in the polar area and persist until two weeks. And suddenly, there is an increase of zonal wind in the tropics. This is zero. So this is an observation. In our forecast model, we also have very similar structure. So the NNO, and after about 10 days, there is a sudden increase of zonal wind in the tropics. So if you look at, because in this region, that's really associated with MGO phase 7. To summarize my talk, this is the second part. So we see there's two interactions between NNO and MGO and the lack of association of the NNO. So I didn't show you this impact on surface air temperature. But really, we have some studies show that this kind of connection impact whether the surface air temperature, there is a very strong signal in North America after MGO. And the NNO in traditional forecast skill is influenced by the MGO. And the MGO skill is influenced by the NNO. So that's really a two-way impact. All right, that's for what?