 All right, can you see that? Perfect, yeah. We'll hear from Amy about the importance of stratosphere in S2S predictions. Thanks, Amy. Great, thank you, Anish. And thank you, Judith, for inviting me to speak today. I'm really excited to be part of this workshop. And today I'm going to be talking about how the stratosphere could be used to predict the weather weeks in advance. I'm going to be focused on the stratospheric polar vortex today, a picture of which is shown in the bottom left here, and how it impacts weather at the surface and how this influences S2S prediction. So I know most of you are well aware of what the stratosphere is. But because we've been focused so much over the past week on the troposphere and ocean processes, I wanted to briefly remind us where the stratosphere is. It's a layer from 20 to 50 kilometers above the surface. And this is well above where our weather happens. And when the stratosphere, early interest in the stratosphere was really not related at all to how it influences weather at the surface, instead, the primary focus was really on the ozone layer, which has its home in the stratosphere and provides protection for us from dangerous UV radiation and allows for life on Earth to exist. And I wanted to bring this up because I'm really going to be focused primarily on dynamics in this talk, not really on chemistry at all. But it's important to remember that ozone interacts with the dynamics. And a lot of the S2S prediction models, for example, don't really incorporate a lot of those feedbacks into them. So that's one thing to think about when I'm talking about this today. So what are the stratosphere troposphere coupling processes that we think might be relevant to S2S prediction? This little schematic shows the general circulation and the features that we think are relevant. And these are little plots that show height versus latitude for the southern hemisphere winter on the left and the northern hemisphere winter on the right. And so here the red blobs are the zonal westerly jets. And so you can see in the troposphere, there's the year-round tropospheric jets. But in the stratosphere, we only have a westerly jet during the winter hemisphere. And in the summer hemisphere, there are easterlies in the stratosphere. And this has important implications for how the stratosphere and the troposphere communicate. And so I'm going to be focused primarily on those jets, which are called the polar vortex. And I did want to point out that there's a lot of other processes that we think are relevant on S2S time scales. I think Yaga will next talk about the QBO quite a bit, so you can learn more about how that might influence S2S predictability. Then there's things like tropical convection, which can influence planetary waves. So ENSO and MJO can generate these large-scale planetary waves that also have a pathway through the stratosphere. And that could potentially extend predictability in remote locations. Obviously ozone could be potentially important, as I mentioned before, particularly in the southern hemisphere spring. And things like solar variability, surface boundary forcing, like sea ice, sea surface temperature, and Eurasian snow cover, anything that can essentially modulate either the mean flow or the planetary waves themselves could potentially drive changes that are predictable in stratosphere tropospheric coupling relationships. OK, so what is the stratospheric polar vortex? This is a nice image of it. It's from February 16, 2020. This is from earth.nullschool.net. And this is kind of an unusual state of the northern hemisphere stratospheric polar vortex. This is when it was really strong. And you can see that it's very symmetric and annular. And a lot of times, the northern hemisphere vortex is much more wavy. But this was kind of a really unusual event that I'll talk a little bit further about later on. But the polar vortex describes the westerly circle polar winds in the winter hemisphere. And these winds are due to seasonal changes in incoming amounts of sunlight. So in the fall, as the sunlight leaves the polar cap, you start to lose radiative heating by the ozone layer in the stratosphere. And so you increase the temperature gradient from the pole to towards the equator. And that drives by thermal wind balance, these westerly winds in both hemispheres. This happens. So this is how a typical metric of the stratospheric polar vortex, which is the zonal mean zonal winds at 60 degrees and 10 hectobascals. And this is showing basically the daily climatology in the JRA 55 reanalysis record from 1958 to 2018. And so you can see the black line shows that daily mean average. And so around January 1st is when the winds on average peak in the northern hemisphere stratosphere. And you can see there's a large amount of variability as shown by the gray shading, which shows the maximum and minimum values in the historical record. And these are what we're particularly interested in, because when we get extremes in the stratospheric polar vortex, that's also when we see the strongest impacts of the surface. And so there's both very strong extremes when the vortex intensifies and becomes much stronger than normal, as is shown in this 2020 image. And there's also times when the normally westerly vortex reverses direction completely. And so the winds become easterly in winter. And these are well-known events called sudden stratospheric warnings. And so I'll be talking a bit about those too. So the important thing when we're talking about stratosphere-troposphere coupling is that it's a two-way interaction. It's not just a one-way direction. And so I wanted to cover both the upward coupling and the downward coupling and how those things happen. And so when we're talking about upward coupling from the troposphere to the stratosphere, a lot of times what we're talking about is how this troposphere generates waves and how those waves can sometimes go into the stratosphere and change the flow there. And so we think that weather systems, for example, blocking patterns and also things like lancy contrasts are associated with these planetary-scale atmospheric waves. And we're talking wave numbers 1 and 2, so barely the largest-scale waves. And if these are in the right location in a way that they constructively interfere with the background waves, they can actually amplify into the stratosphere as long as the background flow is westerly. So there's no way that the waves can travel into the stratosphere, the polar stratosphere, in the summer. So this means that we're really talking about between fall and spring when the stratosphere can receive these signals from the troposphere. There's well-known patterns that are associated with the kind of waves that affect the polar vortex. And so this plot by Garpingel et al 2010 on the bottom left gives an indication of what the 500 millibar heights look like that precede polar vortex weakening events. And so for example, when you see strong wave number 1 type weakening in the stratosphere, you also have sort of a wave number 1 type pattern in the troposphere that's driving that. And this is associated typically with blocking in regions that we normally see blocking patterns occur, such as over the Aleutian region in the Pacific and over Scandinavia for these wave 2 events. Sorry about that. So the vertically propagating waves, which is a positive eddy heat flux or V prime, T prime, can dissipate or break in the stratosphere. And when this happens, it deposits easterly momentum and it slows the westerly polar vortex down, sometimes very rapidly. So the momentum deposition from planetary scale waves is thought to be one of the key mechanisms for driving these disruptions, but it's not the only one. In particular, it's been found that you really need the stratosphere itself to be in the right state in order to break down because you get some internal resonance of the wave once it's in the stratosphere if the stratosphere is in the right configuration. And so these two processes together are able to explain a lot of the reasons why we think these vortex breakdowns occur. And there's usually two ways that these manifest. So I've shown a picture of this here. On the left, we have an inactive polar vortex. So it's a much more symmetric vortex. The black lines here show the potential vorticity contours that give you the shape of the vortex. And when the vortex breaks down in the stratospheric warming event, the vortex typically either displaces, in which case the vortex is basically pushed off the pole over the extra tropics. And you can see the shading here shows the temperature anomaly, so you get a warming that occurs with that. Or the vortex can split into two pieces. And these are particularly dramatic events. And usually part of the vortex will go over North America and Canada. And the other part will go over Eurasia. And when the zonal means zonal winds doesn't just decelerate, but actually reverses direction, we call that a major event. And if you want to look at a review of this, there was a recent review by Mark Baldwin at all in 2021. This past year was published. So all right, so hopefully you can see this little animation. And this is just kind of showing you what is happening in the stratosphere when one of these events happens. The time series at the bottom is you can follow along with how strong the wind was as this event proceeded. This was in 2009. So you can see around January 24th is when the wind actually reverses direction. And the arrows in this plot are showing you the wind direction. So it starts off very westerly. Those are the black arrows. And then by the end here, it actually reverses direction entirely, which is a pretty fascinating thing to see. Because you can see how widespread it is. It's really a hemispheric phenomenon. And you can see that it's really altering the entire circulation in the stratosphere in the northern hemisphere. The black contour here is showing, again, the potential porticity contour. And so in this case, we had sort of elongates first, and then it splits into two pieces. And then the temperatures are shown in the shading. So you can see you get this rapid warming of the stratosphere. All right, so these events are very interesting to watch from a fluid dynamics perspective. But that's not the reason we're really interested in them. We want to understand them because they have a downward influence. And this is shown in this plot using the Northern Annular Mode Index. And this shows a composite of all the historical sudden warmings averaged around the date that they occur. And so we can see that day zero is when the zonal means on a wind reversed. And the NAM responds almost immediately. You see this strong negative NAM pattern, which is the red colors here. And from about 1 millibar to 100 millibars, there's almost an immediate reaction to that disruption of the vortex. But then the interesting part is that you can see around 100 millibars, there's this extension of the anomalies that goes out well into 60 days after the event. And this is because there's longer radiative time scales in the lower stratosphere. And that leads to the persistence of the anomalies. The vortex can't recover there. And so these anomalies just persist for a long time. And that has important implications for using this for S2S prediction. And so you can kind of see that there's downward coupling to the surface. And it's not, it doesn't look as continuous as it does in the stratosphere. There's sort of these drips down. And that's because, of course, there's a lot of other things going on in the troposphere. So we don't always see continuous coupling because there's other factors going on, like ENSO, MJO, that's influencing that coupling down to the surface. One of the interesting features we're trying to understand, actually, is that you might notice that around 500 millibars, there's probably a minimum in the coupling. And then it amplifies, again, at the surface. And this is a very common feature following these events. And we're trying to understand what exactly causes that surface amplification of the signal. So there's not a consensus on the exact mechanism for how the stratosphere influences surface weather. And I'm not going to go into detail on these theories. But I think it's interesting to note that we, you know, there's been a lot of idealized studies on this, a lot of theoretical studies. And we still don't have a really complete answer for how it happens. It's thought that downward control contributes most of the stratosphere response down to the tropopause. And downward control is just where the momentum convergence from the planetary waves slows the winds. And then the next group of waves that comes up has to break at a lower height because they can't travel quite as far up because the flow is slower. But then once we get down to the tropopause, it's not really clear how that interacts with the tropospheric flow. And what is clear is that it seems like you need some, there's some role of eddy feedbacks that contributes to the response that's needed. And I've listed some papers here that you could refer to on that. And then there are some ideas of the remote effects of stratospheric potential vorticity anomalies, which is sort of a newer idea and might explain some of the surface amplification that we see. So I wanted to share this little video real fast. Let me play it. This is from NASA GMAO. And the shading here is showing the potential vorticity. And I like this video for two reasons. The first is that you can see the polar vortex in a very clear way that really describes what it actually is doing. So you can see how fluid it is. It looks, you can really see the wave breaking that's occurring and the filaments that are peeling off. So it's just a really nice way to picture that. But the other cool thing about this video is that red contours here show the 200 hectopascal heights at two different levels. And I think you can kind of see particularly over North America that as this, as the vortex starts to split, there's some nudging of the troposphere below it, the jet stream below it. And so this hints at sort of how the stratosphere can potentially interact with the tropospheric flow. And so what this does to surface weather is shown on this top left plot. This is a composite of the mean sea level pressure anomaly, the surface temperature anomaly and the precipitation anomaly, day zero to 60 after historical sun warmings. And so you can see that there's a really nice negative NaO type pattern in this mean sea level pressure. And this is associated with very colds, air outbreaks over most of the extra tropics and also anomalous warmth over Greenland and subtropical Asia and Africa. And additionally, there's precipitation anomalies particularly over the North Atlantic and European region. And so you have this thing where you have a weaker vortex, you actually get these really cold mid-latitude extremes and snowier weather and some anomalous warmth over Greenland and Asia. On the other hand, and this is shown by the schematic in the bottom right, when you have a really stable or strong polar vortex, all that cold air over the Arctic is contained. And so you end up with a warmer winter weather over the Eastern USA, Europe and Asia and drier over Southern Europe. And so these changes are persistent and thus potentially predictable. So here's some examples. This was a sudden warming event in 2018 that's very well known. In fact, it had such big impacts in Europe that they named the storm, Beast from the East. But there was a number of storms, not just that one that occurred following this sudden warming. There were also about four Nor'easters that March that came across the North Atlantic due to the strong blocking pattern that was over Greenland at the time associated with the negative NaO following this. So that Northern Hemisphere isn't the only one influenced by the polar vortex. The Southern Hemisphere has a polar vortex and it can also have impacts. I should note that major midwinter SSWs are rare in the Southern Hemisphere. There's just not as much planetary scale wave forcing. And so the vortex is much stronger. It's a lot harder to get a reversal of vortex. Instead, there's a lot of variability linked to surface coupling, which occurs in Austral Spring. But this was one example in 2019 where there really was some significant impacts. And Umpal Lim has studied this and she has a paper in Nature Geoscience showing that when you have even an anomalous weakening of the polar vortex, you get increased chances of exceeding extreme maximum values of temperature, rainfall and wildfire danger over Eastern Australia following these events. And a prime example of this occurred in 2019 when we actually had a near major warming. It didn't quite reverse, but it was extreme deceleration of the vortex. And then it was followed by the most negative Southern annular mode on record. And obviously that summer in the Southern Hemisphere was followed by some of the worst wildfires they've seen. And obviously had big impacts. And we had kind of picked apart this relationship in the 2021 BAMS paper that just came out on this event and show that the vortex breakdown contributed about as much as the record strong Indian Ocean Dipole event. So it can be a major player in these extreme. So Daniela Domais and I highlighted this in a recent paper to Nature Communications on how the stratosphere drives extreme events at the Earth's surface. And basically one of our points is that, when we typically hear about the polar vortex, we all think cold, snowy weather, but really a sudden warming can lead to impacts in a variety of different ways, including things like heat over Africa and Asia, flooding events, storm series over the Northern Atlantic. And in the Southern Hemisphere can also have impacts like wet spells over South America, drought and wildfires. And these things can further have impacts on health and transportation, energy and agriculture. So I wanna point out that it's not just a breakdown of the polar vortex, but also a very strong polar vortex can drive surface extremes. And I hinted at this at the beginning of the talk, but in 2020, we saw this really strong polar vortex which is shown here in this plot by Zach Lawrence, which shows the 10 millibar zonal means zonal winds as a function of latitude. And you can see it's very exceptionally strong. And the bottom plot shows this as a function of pressure and time. And you can see as far back as December 2019 at one millibar, there was some acceleration of the winds. And this really continued for an extremely long period of time. It was very persistent. And of course, associated with that was the strongest Arctic Oscillation or Northern annual mode in the 70 year record, which explained about two thirds of the warmth that was then observed in many over Russia and Asia. It was the record hottest winter in many locations. I'm sure in part due to climate change but a large part of that was just this climate variability and this strong coupling between the stratosphere and the troposphere. So now I wanna talk briefly about what we're doing to analyze these relationships in the S2S models. And so we have a group called the Stratospheric Network for the Assessment of Predictability or SNAP. And this is both a WCRP spark international activity and also it's a part of, it's a sub project of the S2S prediction project. And so I'm one of the activity leaders with Higham Garfin goal and this shows the committee and we're always welcome to have new, especially early career people come help analyze the S2S models and the stratosphere. So if you wanna join us, please let me know. But our goal is to assess the stratospheric predictability and its tropospheric impact. And so we recently did a few studies on this and one of the things we noted is that, well, most of the S2S prediction systems now do have high model lids and are more vertically resolved above a hundred hectobascals. And so this plot on the left shows which of the S2S models have various levels in the stratosphere and their model lid height. And so we decided any model that had a lid above 0.1 would be considered a high top model and most of those also have more levels in the stratosphere. And then we have low top models that don't have as high of lid or very many models in the stratosphere. And then we analyzed some features in the S2S models in terms of both stratospheric predictability and its role on the troposphere. And so this plot is, you don't have to understand the details of this but the main point is that the stratosphere has longer memory than the troposphere. And this is particularly indicated by the numbers at the bottom here. And so this is showing the anomaly correlation at 50 millibars and at 500 millibars from 20 to 90 North in the Northern Hemisphere but similar results are seen in the Southern Hemisphere. And so here we were evaluating at what point does the skill fall below 0.6? And so at 50 millibars, it's about double that at 500 millibars. And this is true surprisingly even in the summertime. It's only a difference of about three days in that case but compared to eight days in the winter time. And then we correlated this with, so we correlated the prediction scale at 50 with the correlation scale at 500. And we found that models with longer prediction skill in the stratosphere have longer prediction skill in the troposphere. And if you consider only the models with a high top or a better resolved stratosphere, those also tended to have both the highest correlation in the troposphere and in the stratosphere. So of course the direction of causality here can't be inferred because if you have a model with a good troposphere, it most likely also has going to have a better stratosphere. It's not necessarily saying the stratosphere is driving a better troposphere. So we wanted to further understand can we look at forecast of opportunity when there's an actual event going on in the stratosphere and see if that changes predictability in the troposphere. So before I get to those results, I wanted to point out that we also considered the predictability of the vortex itself and these extreme events in the stratosphere. And so we looked at strong vortex events and sudden warming events. And so this shows the percent of ensemble members capturing observed events during the S2S record as a function of lead time. And so you can see obviously the more and more members capture it as you get closer to the event, but you can see that you only get accurate detection or more than about 75% of ensemble members at 10 days or less on average. So some events you get a little longer lead time, but in general, these are not highly predictable and that kind of makes sense because they're driven by things like weather patterns in the troposphere which themselves are only predictable at 10 days or less. So unfortunately, it's really hard to see a lot of these polar vortex extremes far in advance. It is true that models with a higher model top detect events at longer lead time. So there is some gain to improving the stratospheric representation. Okay, so we wanted to look at the surface response following both weak vortex and strong vortex events in these models. This shows the week three, four temperature anomalies following polar vortex extremes. And to do this, we looked at the time of initialization whether the zonal mean zonal winds at 10 millibars and 60 North were either weak, which was considered less than five meters per second or strong, which was greater than 40 meters per second. And first we just made a composite of what the response was at week three, four. And so the plots on the far left show the error interim response during the S2S time period that we're considering for both weak and strong vortex events. So you can see they are basically more or less linear, although there are some differences in some regions between the strong and the weak events. And then we consider that, we compare that to the multi-model mean on the right. And so you can see that in general, it's really capturing that NAO sort of pattern in the temperatures that we would expect following these events. So there's some differences that we should note. For example, in, there's warmer anomalies and error interim compared to the multi-model mean. It doesn't capture the strength of some of those warm anomalies over Greenland or during the strong events over Eurasia. But in general, the models are doing a pretty reasonable job. Okay, so then if we look at the skill of the forecasts afterwards to do this, we really needed something to compare against. And so we created control forecasts which used the same dates as all of our weak and strong vortex events but had randomized years. And so then we were able to get a set of forecasts where there was presumably nothing going on in the stratosphere to compare to. And so that's what's shown on the left here, the correlation and the root mean square error. And then the middle plots show what happens following these weak vortex events. And so we clearly see some increases in correlation skill following these events, particularly over Eastern Asia, the subtropical Asia region and parts of North America. But what we found that was surprising is that notably there's a decrease in correlation skill over Europe relative to the control forecast. And this is pretty surprising and somewhat unfortunate because obviously a big source of potential predictability for Europe is stratospheric variability. We think what's going on probably is that because temperature at two meters in particular is very sensitive to exactly where the jet stream moves, that these weak events can actually drive increased variability of the jet stream over the North Atlantic. And so it's actually quite a bit harder to predict the temperature changes after the weak vortex events in that region. And so that's why we're getting a decrease in the skill there. The root mean square error mostly decreases everywhere for forecasts, so that's a good thing. There's less error in the weak forecast relative to the control in most locations. So this just kind of summarizes the changes in skill over two meter temperature following the polar vortex. And here the error bars are showing using a bootstrapping technique, how significant those changes in skill are. So obviously if we look at the Northern Hemisphere, there's weak increases in skill, but it's not very significant. But in certain regions there are significant increases in skill and then notably over Europe, we see less skill following these events. Amy, a couple of minutes. Okay, I'll tie up. One thing we think is that large-scale circulation metrics could be more useful than something like two meter temperature. And so this is showing the Northern Annular Mode index in the same sort of way I just showed for two meter temperature at both 100 millibars and 1,000 millibars. So this is like the lower stratosphere. So clearly there's a nice increase in skill at 100 millibars following both strong and weak vortex events. But then at the surface, it's noisier, but in general, there's clear increases in skill for almost all the S2S forecast models. And so the idea here is that something like the NAM is better predicted than inferring the two meter temp through a statistical approach during post-processing may be a better forecast than trying to predict two meter temperature directly. And this is an idea that was also presented in Skype at all 2014. I'm gonna skip this part since I'm running out of time, but I wanted to quick touch on some issues that I think are related to stratosphere-troposphere coupling in S2S prediction systems. As in many other parts of the climate system, there's systematic biases that prevail even as the systems raise the model top and increase vertical resolution. And another issue is that the timescale influence varies a lot from event to event. That's shown in this plot on the left, which shows two different stratospheric warming events. And then I've kind of highlighted where week three, four occurs in that. And obviously during 2018, during week three, four, there was really nice downward coupling. But in 2019, the downward coupling occurred after week three, four. You can see there really wasn't anything going on during week three, four. So how do we best capitalize on a stratospheric windows of opportunity when there's so much variability on when that happens after the event? This ties into the fact that the surface response to any given polar vortex event is highly variable. Obviously there's a lot else going on. There's the role of tropospheric influences and large internal variability. And finally, there's still missing processes in the S2S forecast models. And I think Yago will touch more on the QBO, but also ozone feedbacks. There's really not ozone chemistry in these models. And I guess the question is, does there need to be and what's the best way to do that? Can this be improved? All right, so I hope I've shown that the stratospheric information and particularly from the polar vortex is useful at a minimum during windows of opportunity for improving predictive skill on S2S time scales. But there are inherent limitations in regions where winter weather can actually be nominated by these events for weeks. But the event itself is only predictable at 10 to 15 daily times. There is potential for probabilistic forecasts of stratospheric polar vortex events at longer leads. I didn't get to touch on that, but more work remains to determine how much skill can be gained. And then how best to utilize and convey this information in real-time S2S forecast is still a real question. At this point, talking to people at Climate Prediction Center, for example, they're very aware when these events occur, but how to use that information in real-time can be a challenge, I think, especially because they don't occur that often. So with that, I'd like to thank you. And here's my email and my Twitter handle. If any of you wanna follow me on Twitter, I do post a lot about the polar vortex, so you can follow me there. Thank you. Thank you, Amy.