 Our next speaker, the Magdalena Balmaseda. Magdalena is the head of the Earth System Predictability section at ECMWF, and also gave a lecture at our student colloquium a couple of weeks back. Thanks for that, Magdalena, and thank you. OK, thank you, Anish and Judith, for giving me the opportunity to present some work from ECMWF. Let me share my screen. Today I'm going to talk about some recent work. This time it's on seasonal forecast rather than subsistional. And it's on the impact of recent Indian Ocean variability on seasonal forecasts. There are two specific topics. One is the Indian Ocean impact on the two-year El Nino of 2014, 15, 15, and 16, the implications for the two-year predictability. This is some research that has been published. The other is in preparation by Redis Senan and is the recent impact of the Indian Ocean SST anomalies on the positive El Nino of DJF in 2019 and 2020. So I start with the first one. If for whatever the reason, I get cut, please turn me, because my Wi-Fi is not so good. And I'm going to switch off my camera. Yeah, so let me go back to the presentation. So we had a look at the big El Nino of 2015 and 16, and we compared it with the 97-98, that it was the previous larger El Nino in terms of sea surface temperature anomaly. And we had a look at the energy budget. So during El Nino, we know that the Ocean Pacific discharged its heat. Quite a lot of heat goes to the atmosphere, as we can see here, for 97-98. The net balance of ocean heat content before and after El Nino in 97-98, it was quite a large discharge. I cannot remember the use, but sorry, realize that they are not there. And well, that was just a diagram of what is the energy cycle for 97-98. And in contrast for 2015 and 16, the Pacific Ocean, the Tropical Pacific, didn't discharge the continued quite well charge of heat. And the impact of the atmosphere was substantially smaller. So it didn't release so much energy into the atmosphere. But there was a big difference between the two events that it was due to the Indonesian through-flow. You know that during the El Nino, the Indonesian through-flow weakens, and there is not so much heat export into the Indian oceans. If we talk anomalies, this is equivalent to anomalous import of heat. And that was extremely large in 2015 and 16. And you can see it in terms of time series from ocean re-analysis that the volume and heat fluxes associated with Indonesian through-flow were really unprecedented. It's very weak. And this also can be seen in the strength of the Indonesian through-flow. It's proportional to the sea level gradients between the Indian Ocean and the Pacific. So we started investigating why it was, what happens, why was the Indian Ocean, the Indian Indonesian through-flow so weak during this year? So those are the questions that we were trying to answer. Was this weakening predictable, the Indonesian through-flow transport predictable at internal time scales? What was the role of the Indian Ocean state on this weakening? Did the Indian Ocean state influence the weak El Nino of 2014 and 15 and the strong of 2015? And does the Indian Ocean state influence the predictability of ensuing the two in the second year more generally? So for that to answer these questions, we set up a seasonal forecasting experiment based on system five, the operational system up to 24 months, starting the 1st of February, 2014. And the experiments consisted on swapping the Indian Ocean initial conditions, replacing the ones in 2014 by the ones in 1997 and combinations of the kind. And we launched several 50 ensemble members. So these are the preamble to see whether the methodology will work. The first thing that we can see on the right-hand side is the Indonesian through-flow. We see the black line is the analysis accumulated over the two years. And we see that the seasonal forecast indeed captured this weakening. I mean, positive values means towards the Pacific. When we replace the Indian Ocean by components of 97, 98 Indian Ocean, this Indian Indonesian through-flow doesn't work. So there is indication that the Indian Ocean matters and is predictable as well. And then the other thing was what was the impact of that on the Indian Ocean here? We see how Miller diagrams. This is the observed anomalies from the analysis, 97 minus 2014 in sea surface temperature, keep content, solar winds and velocity potential across the equator. And you can see that these two events, 97 was much stronger than 2014. And it was followed by La Nina, while in 2014, I mean, 2015 was a very big enemy. That was a big difference. And you can see it in sea surface temperature and you see this huge propagation of a negative anomaly of heat content as well. And then different feedbacks, the bernet feedback as well. So in our experiments, seasonal forecast experiments, if we take the forecast in 97 minus the forecast in 2014, we see that they capture to certain extent the ensemble mean captures this strength, stronger warming in year one and stronger cooling in year two. And now we go to the swap experiments. And here what I'm showing is the difference, the impact of the experiments where we put the Indian Ocean initial condition from 97 minus the reference. Everything else is like in 2014. And you can see that in fact, the Indian Ocean in 97 would have produced a much warmer ENSO in year one and it would have produced a colder event in year two. And you can also see the indication of some Indian Ocean diaper developing in the autumn of the first year. You can also see that there is quite a large signal in the second year coming in the ocean heat content. If we just change the Indian Ocean in the surface in the upper 20 meters, we see that some impact just coming from the atmospheric bridge. But we see that impact and that's visible in the first year but it's not so visible in the second year. So looking at extremes, what we looked is classified in different types of events, extra strong and Nino moderate warming, neutral or La Nina in the, for year one and year two and in 2014, we saw that most of the ensemble members predicted moderate and Nino with some extreme. That contrasted very much with the 97, 98 where the probability of very extreme and Nino was quite high. And as we see the probability of very strong La Nina. When we show up the ocean, the Indian Ocean, what we see is the probability of the warm event with 97, 98 Indian Ocean, it increased the probability of extreme and Nino at also increased the probability of La Nina in the second year. So this is a clear indication that the Indian Ocean affects the predictability both years of extreme event. And these other diagrams, what we try to say is to condition that of the strength of the Indonesian through flow. The Indonesian through flow is weak. No, it's a strong with the 97, 98. And the ones that they have a strong ITF, they tend to produce La Nina in the second year as we can see here. And with the weak ITF, the PDF is shifted to the right towards warm conditions in the second year. And this Indonesian through flow, I'm going to go quite quickly here, this difference, this weakening is clearly different from the internal variability from the impact of El Nino. So the difference that we have cannot be attributed to the impact of El Nino. It has to be something else. So the questions are raising this, what caused that unprecedented weakening? And one possibility is the preceding PDO event because it's consistent with this in each of the sea level gradient across the maritime continent. So in the preceding PDO, the winds will have and raise the sea level in the Indian Ocean in favor of weakening of the through flow. So what are the prospects of the predictability of ENSO GR2 for two years time scale? And what are the windows of opportunities that have that change related to changes in the PDO? There is also the implication for interpreting the seasonal forecast. As you know, in 2014, we were very eager to announce a big El Nino coming, but actually the forecast was quite probabilistic. Had we had the second year, the two years forecast and had we seen the probability of extreme El Nino was already there in for the second year and not probability of La Niña, maybe we had issued a different, interpreted the forecast different. And so that's one part. The second part it has to do with this positive NEO over that affected the weather over Europe in 2019. 20 DJF, you can see here in era five. I show seasonal forecast, but it was most of the seasonal forecasting model predicted this very well, which is an unusual occurrence because the NEO usually is poorly predicted. So we try to say, okay, what we had the option repeat the same experiments, but not with the couple model, but with observed SSDs. We also have similar results. And this uncoupled experimentation will allow us to answer some questions, what was the role of the sea surface temperature anomalies in different basins. You can see here that in 2019 and 20, there were very strong SSD anomalies in the Indian Ocean. And that were very well predicted. So we did that. There are several experiments here. I only show the results from two. In the left column, it shows the impact of the, when we have the system type with observed SSD, another experiment where we replace the Indian Ocean with climatological values. And in this case, in the right column, is when we replace the Pacific Ocean with climatological values. 200 kPa, merional winds, 200 kPa geopotential height, and the Rosby wave surface. So you can see that the NEO footprint is much stronger. It's a strong associated with the Indian Ocean SSD anomalies. So the results are consistent with a yield type response to heating, tropical heating, plus westward propagation of Rosby waves. So this westward propagation is a little bit not so common because traditionally we only think on Rosby wave propagating eastward. If that's the case, then they have to travel a long way before reaching Europe. And that's probably that was the reason for the low predictability. But several studies among them, one that explains it very well is Shaman and Superman in 2016, sorry, is that if you have not sonally symmetric flow, but sonally asymmetric flow, and with very low sonal wave numbers, this westward propagation is possible, especially on the North Africa and Egypt, because you have very high values of potential vertical gradients and the sonal flow is not so strong. That means that the geographic proximity of the Indian Ocean to Europe, it will favor the predictability and the signal to noise ratio. Further experiments show that this response is modulated by the amplitude of the Indian Ocean SSD, and also by modulated by ENSO. So in 2018, there was no ENSO and that favored the strong response of a year. Just a couple more minutes. Yeah, I just finished after showing these results. And these two examples, to my mind, they show that the variability of the Indian Ocean has serious implications for seasonal forecasts. And both the weak ITF in 2014 and 16 was responsible for the two-year warm event that was likely predictable. One conjecture is that this weak ITF was induced by the preceding PDO phase. It could be by climate change, but this is a question, it's an invitation to opinions. The other one was that unprecedented high seasonal predictability of the NAO in DJF is attributed to the Indian Ocean anomalies. If this SSD anomalies that they are linked to climate change continue to raise, it may have implication for seasonal predictability of the Europe. Both cases have implications for the design and interpretation of seasonal forecast system. So how to extract information from a seasonal forecast in the context of a climate change in climate? A first option, that's what we do now, is to have long enough records of pre-forecast. But even if we could afford it, would this be enough? Would it be a good assessment of a scale in a non-stationary climate? Because we may see next year, we may see the unseen. So do we need to have more insight into the processes leading to, involved in the predictions? And the second question is, do you think that seasonal outlooks, offensive at the second year would be useful? And that's it. Thank you very much for your attention. Thanks a lot. Again, it was really interesting impacts of the Indian Ocean. Any questions for Magdalena? So I had one, Magdalena. You mentioned about the climate change aspect as well and how it would impact future prediction, right? And the Indian Ocean seems to be the region where the warming, at least as per historical records, the warming is more than other regions and other global oceans. And how much would this trend be a source of predictability versus it would make the scale worse in the future if it continues this way and the Indian Ocean continues to warm more than the Pacific, for instance? Would that help improve the seasonal prediction because there is the trend and maybe the model can pick up on the trend or would it make it more difficult with the Indian nation through the flow of changing? I mean, I don't know. I mean, first of all, I mean, I made the link with climate change because it's the gut feeling, but this has to be demonstrated indeed. So I think there is a lot of research to do. So there are two aspects regarding the Indian Ocean impact over Europe. The response seems to be quite... If you... Depends on the amplitude. But imagine that if the Indian Ocean is to continue to rise, the only thing is that they may not rise out in thin item. So how? How? So, but there will be a saturation point, but it will have implications. So I think we need to understand. But it's true that there are contrasting impacts because on one hand you have the Indian Ocean warming, but also the Indian Ocean warming could be more warming in the Pacific, the other Indonesian through flow. So maybe there are competing impacts. And here in this case, they were very isolated cases. If we put the two together, I don't know. Yeah, that's a good question. Great. Thanks, Matin. Jan, you have a question in the chat. Yeah, thanks, Matin, for this great talk. Very impressive results. I was wondering the results you showed with the winter 1920 and the higher predictability for the NAO. I remember that there was also a strong polar vortex in the stratosphere. And so I was wondering if you looked at the influence of that on the overall predictability of the NAO and especially if the combination to the results you showed, because there was an influence on that. We have looked a little bit. So the Indian Ocean seems to have an impact on the strengthening of the polar vortex. But in our predictions, at least in our system, those are largely canceled by the impact of the Pacific. So the polar vortex and the eastward propagation of Rosby waves in our system and the Pacific and the Indian Ocean cancel each other. More or less. So the source of what? Mostly. I mean, there is a little bit of eastward propagation of Rosby wave, but the sources of potential of Rosby wave at the exit of the jet, they are in the opposite direction, both of them. And the same with the polar vortex. The outcome is that the Pacific influences dominates. So probably the polar vortex doesn't strengthen as much in normal model as in observations, but it's huge spread by the way. So we don't know whether it's predictable. Thank you very much for that very good question. Yeah, thanks for explaining it. Thanks, yeah. And thanks again, Mike. I was really fascinating talking. You findings with seasonal prediction and yeah.