 This is Rich Neal. Rich Neal is the head of the atmospheric modeling and predictability section here at NCAR. He is working, he's also the co-chair of the atmospheric modeling working group, oh maybe no longer. Your bio is outdated. He does a lot of development with the community system model. The atmospheric component has done a lot of work on the MJO, on blocking and on making the atmospheric component better. Introducing vertical levels, whatever gets to the skill of CSM2, Rich had his hands in it. So Rich, go ahead. I'm looking forward to your talk. All right, thanks Judith. I shall share my screen now. Can you see that? Perfect. It's not yet full screen? Yeah. How about now? Yeah. All right. Well, it's nice to be you with everyone here. I know the ASP colloquium is fun. I've been involved in the past and it's a lot of hard work. So thank you Judith and Anish. Okay, so today I'm going to talk predominantly about atmospheric blocking, a little bit about teleconnections and some about the systematic biases that cause problems. So this is work with the help of these sort of fine people here. And this is my kind of very brief cliff notes of what I will go through. So, you know, in essence, I'm not sure of people too familiar with aspects of blocking. So I'll take some time to go through and why it's important. So blocking, you know, the easiest thing here is just to define it in terms of meteorology. And the key things are, you know, the nearly stationary patterns, high pressure at the surface effectively. And what they tend to do is redirect the kind of more synoptic typical circulation patterns in the northern hemisphere to the northern side. And so this is the only time I talk about the US just because it's a nice example of different types of blocks. So you've probably heard, you know, you think about, especially recently in the northwest there, which is probably a Rex block, I haven't looked in detail, but basically these kind of stationary block patterns to be of a number of different types. So that's just to orient you kind of geographically what blocks are. And of course, you know, in terms of impact and synoptics, Western Europe is a big region for blocking and the impacts of blocking and just a couple of wintertime, you know, examples here from the UK. So this is kind of the first time I've ever seen the whole of the UK covered by snow and Ireland still manages to escape it the Emerald Isle. And over Western Europe here during, you know, 2009, there's a very extensive cooling at the surface, surface temperature, and this is associated with the black. And also in summertime, probably hear more about the impacts of blocks in summertime, so very stationary high pressures over these particular regions. Again, you know, 2003 over France, I'll talk about that in the next slide, was very hot for a protected period, so 2006. And so on the right here, thinking about, you know, Western Russia on the right, this is a time series over Moscow through that period several months here. And this is a metric of blocking. And I'll come to this later. But essentially, this metric of blocking, you know, was stationary for 45 days. This is kind of the ultimate impact was these real severe up to 12 Kelvin anomalies during that period. So big weather impacts. In terms of further impacts, you know, because blocking weather, etc. And then what are the societal impacts. And this is a really nice example of what you need to get right with blocking to think about getting the right societal impact. So just go through this. So the sidelines here are temperature and blue, it's like climatological through summer and then red was this particular period in 2003. So we had this protracted period of, you know, over 30 Celsius. And the key thing to note from this, if you look at like mortality and blue is kind of standard climate climatology of mortality and such thing exists. And so the key thing was blocking has to exist for several days before, you know, the impacts, you know, on your health, etc. are seen to the extreme. So that's the key thing, block it. It's not a question of just capturing the blocking, it's capturing it for extended period. And so in terms of the impact of blocking, it relates from the bottom right here to persistence, having a clear sky gives you surface heating or high temperatures in summer and stability, high stability gives the rest of the quality. So just just a summary of how all these kind of kind of conjoined to give the strongest impacts. And, you know, you can choose your, you know, interesting one here in terms of the, you know, the knock on aspects, a lot of impacts on crops, for example, but wine was good. Wine was better during this year. There you go. So globally, we think about what how to categorize and what synoptically does it look like it tend blocking tends to occur is 2006 in this case, as a more global pattern. So it's not it tends not to just be isolated on the whole. There are tends to be, as you can see, you know, easily, you could easily envisage the wave pattern of, you know, warm, cold, warm, cold, see the temperature and all these. And if you look at that from the perspective of the circulation on the left here, for that particular period in the previous slide late July, for example, in climatology, even in summer, you get a pretty much, you know, to the south of its winter position, zonal jet that is pretty consistently zonal. But for that particular period of what we saw, there was a very large excursions both to the north and south that is this block, you know, gives rise to high pressure, low pressure, climatology, because so it's important to categorize that with respect to trying to predict blocking of course, blocking itself is really tough, because if you think about the opposite phase baroclinic circulations, there are a lot of theories about baroclinic wave growth, how fronts grow with agiastrophic theory, for example. And it's, you can, we tend to and most modeling centers do are able to run simplified models that see how well they reproduce baroclinic activity and they do on our very well. But there's no such test for blocking. There's no, there's no kind of canonical way to think about blocking per se every event is kind of different. And a lot of these events associated on the right here, this is a pattern of vorticity, relative vorticity. Yeah. And so basically in the, in the, in the grace here, it's high pressure, so to answer climate circulation. And what tends to happen is that, you know, there's folding of the interface between the high and low vorticity and these tend to be wave breaking events. So that's just either cyclonic in this case, I think, yeah, cycle, no, I'm cycling, and then cyclonic wave breaking. And so that's the kind of key thing about whether, you know, you locally you have a jet or whether the wave breaking of the exit of the jet gives rise to these blocking type events, but they're tough to predict. I mean, phenomenologically, what do they like? How can you categorize this? It is tricky. And people use many methods to categorize this. This is kind of four ways, some simple, some more complex ways to categorize this. So this is an event in 1994. And this is from era interim analysis. And so what we see here is, you know, in the top left, simply surface pressure for this period, obviously high pressure, clear, that's a way to categorize this. Also, what we will look at in terms of, this is the most standard metric is the 500 millibar height. And it will turn out that a good metric is the result, the meridial or graded to that quantity. Also, so on potential vorticity is conserved on theta surfaces. So this kind of categorize, you know, the column PV, which you put obviously to the stability and vorticity and that's conserved. So that's pulled off from the higher PV regions into the lower, essentially, climatologically lower regions in that category as well. And also, you know, there's a reversal of the wind. So there's easterlies. And that's typical of Western Europe. Easterlies, cold in winter tends to be warmer in winter off the hot continent. So that's just to give you a sense of, you know, the categorization. And if, you know, I'm not sure whether you can see this over Zoom, but you can see it over that, you know, the period of 10 days is this particular block is evolves. At the end of the, if we look at there, bottom right here, the jet is extended there. And then we can see that it's cut off with those Easterlies. So these Easterlies come in and cut off the jet. So that's just to give you a sense of blocking. Okay, just summary blocking there. In terms of prediction, in terms of S to S prediction, it's very hard to predict blocking. And it's only in the last few years that there's some progress that has been made. But it's also a hard and a standard, you know, up to 10 day forecast. And so this is a really nice example of them involved now. But on the left here, you can see the position of the 500 millibar height in here, 33 members. So 33 members where this 500 millibar height is this particular value. They all agree in the Atlantic here because it's barricade. So there's a strong jet barricade wave both region and they tend to agree in this particular region. But when it comes to what happened was there was a block in Eastern Europe here. And we can see that there is disagreement. So some of them on some of our members keep this 500 millibar height level to the south, but a lot of it keep it to the north, which is then becomes a disagreement, bifurcation, if you like, of a block versus barricade continues. And it can give you that in my simple minded thing is that barricade sits across this kind of left detractor here and you're always in extension. But if you initialize the forecast close to this connection point between two attractors, then you know, problems happen. So you can get bifurcation. And another way to see that is just the lead time forecast here. This is a period here. This is in London, I believe. Yeah. So a prediction of London, 10 day forecast, 30 round time on members, deterministic and analysis. And yeah, at least to day six, they're all in good agreement. This is surface temperature. So in those kind of scenarios, regimes are predicted. But of course, in blocks, similar to this kind of scenario here, by day two, there's a lot of is a lot of disagreement among among the members. And even, you know, the deterministic one, the high, high resolution one is, you know, is of order six Kelvin wrong along with the many of the ensembles by day two and a half through. So this is a key example where you really need to understand the spread and and and span that spread with a number of ensembles. But basically, you know, we're thinking about S to S prediction of blocking, but like we're at day three here, so you can see the challenge. So in terms of modeling, again, back to measuring, not just modeling, but categorizations of blocking. Here we have a couple of measures of blocking, one dimensional one, a simpler one, and a more kind of detailed two dimensional blocking. But as I mentioned on that, on those four kind of ways to categorize blocking, this is a 500 mV height gradient, essentially, in in the zonal region. And it, there are two measures, there are two metrics of measures of blocking. And there is a criteria that the difference between these two measures has to exceed a certain value, which would then essentially categorize this folding of the z 500, which looks like a block. So once you, once you exceed that, then you can add that to your proportion of blocking for that period. I'll come to the two dimension ones next. So if you do that, what you get is, you know, one dimensional line plot here. So we're looking at, just remind myself around, I forget the, yeah, 15 North, essentially 15 North is the center point of this. So what you get is a distribution of blocking frequency. And the black lines here, a couple of reanalysis, and so let's focus on those first. And we can see that for this, you know, this period is 1790-2005, two very obvious peaks, Western Europe, as we talked about, but also the Northern Pacific, Central and Northern Pacific. And these are around 20 odd percent of the time. So according to this metric, you know, we're at 20, 20 percent of the time is blocked, and presumably the 80 percent is more barricade. So it's a lot in these peak regions. And then there are a minimum, essentially over, you know, there is a minimum really near the US and then central Russia. But that's not to say the blocking does not exist. It's just by this metric, there's a minimum. This is wintertime, sorry if I haven't mentioned. And the key thing to take away, two or three things, is these are CMIT models, the last assessment five. And so there is a ton of variability. Climate models do not do a very good job of capturing the frequency of blocking by this metric. The second thing is, you know, the key region of Western Europe, where I showed all these examples, all the models, you know, some are close, but all the models do not, none of the models capture the correct frequency of blocking. And that's a theme of what I show next. A couple of models here go close and there are actually a lot, there are high resolution, sorry here, there are high resolution GFDR models, but of course, not of course, but consequences it overestimates into the continent. So let's just emphasize that high resolution does not necessarily help you. Because as Judith said, my interest is CSM and CAM. And so if we look at, you know, three versions of CAM maps across to the block, the blocking seems the same as what we showed before, we can see that in wintertime, we have this problem. No matter the version, I'm not quite sure of CSM too. I do have that plot, but CSM one has some improvements in DGF, but we really struggled to capture this particular, you know, Western Europe one. This is secondary Greenland plot maximum, we also failed to capture that. You know, but in the Pacific, we're doing a pretty good job. The variability is higher in the Pacific, and that's probably related over this time to Enso. But in general, the client model does a good job, but just to emphasize all models do not do a good job in Western Europe, which is a problem when you think you're using them, particularly in S2S framework, using climate models with initialized states from analyses. It's hard to capture that. You know, you can look at other seasons where blocking is slightly less important. March, April, May, we do a good job. The jet is weaker. So it may, as I'll show in a second, may relate to the strength of the jet that is really making it hard for us to reproduce. And we've made some improvements. And also in summertime here, much weaker frequencies, but still, still work. So that's just kind of an aside of how we do. And as I mentioned, the jet, we can see here, now I've split it up in terms of the blocking strength for a particular longitude. So I'm doing this analysis along all these longitudes for a climatological year. So every time it's orange or hatched, then that's blocking. And every time it looks kind of blue, then that's more broken. So you can kind of pick out the improvements that I talked about. Canthru was really poor for most of the time in Western Europe here. And we've clearly made some improvements through to Canthru. So I mean, you can see this blocking is not stationary, sometimes it retrogrades with time, sometimes it moves further to the west. So there's a lot of characteristics of the blocking that is really hard to capture. And just coming back to the strength of the jet, maybe responsible for some of these biases. And this is the etiquette energy of the jet in wintertime. This is in Canthru, which does, I don't show it next to observations, but it does a very good job. And if I look at Canthru, then the strength of this jet really is way too strong, the energetics of the jet, well into Western Europe. And this might explain why. This rosby weight breaking just cannot, is either too far east or more likely, it just cannot break down the jet through blocking processes. Okay, what's next? So, to now to combine the model using it in now, we're using the prediction mode. So these caps, which are similar to S2S type simulations. And again, we're predicting, just a little bit of information here, but 2008, 2009, so the winter of that, part of this year of tropical convention period. This is CAM5, so it's not the most recent version of the model, but some S2S has been done with this, I think. And so we're looking through the season where it's been some blocking, based on that metric. And, you know, solid line, dotted line, that is day zero, so the analysis essentially. And so just the dotted line. And we can see the evolution over just five days. So we're not really into the S2S period. And so we can see that, you know, the biases of the model come out pretty darn quickly. Where we said it was a problem. So DGF, it reduces its magnitude in the western Europe. And then it increases the magnitude further to the west. So although it's not necessarily decreasing everywhere in that region, we can see that that is the source of the greatest biases after day five. So, and then it seems to be continuing to get worse the one day five. This is seen a little bit in summer where we do see a reduction with that. But it's just stark to see that the biases of the model in climate mode really are the places for blocking where they really come out in these shorter term forecasts. So again, bottom line is the biases in the model really are really going to matter because they matter even after five days. So you can imagine after 20, 30 days for sub-seasonal. So I jump now to the two-dimensional metric. So this is essentially kind of a repeat of what is done here. But now we're doing it essentially in two dimensions, not just doing this metric. You can do it a number threshold of course. And you can vary this, but this is the way we did it in this particular way. And so this is a 2D metric. So you're going to see some geographic plots of blocking. So this is blocking with respect to wintertime here. And so there's a little bit more detail now. So essentially, although the blocking was specific, we seem just as strong in some sense in this region compared to the Atlantic and that's very specific. It's a little bit stronger but at a higher level. And now we can see that the green is blocking much more obviously. So these are the focus reasons we're doing focusing on in previous research to try and understand the biases here. And just for interest, there is blocking that occurs in the southern hemisphere that Australian researchers are interested because it relates to the southern impact of El Nino for example. How well do we do? Again, just a quick plot of things that have improved. And again, DGF, no surprise based on even on the 1D metric blocking, you know, Western Pacific really is the killer region for, that's sorry, Western, my region's mixed up, Western Europe is really the challenging region for blocking, which is unfortunate as I said because that is where we think about blocking and the NAO, North Atlantic Oscillation, trying to forget that, this is going to be a tough model to do that. Seems like there is some improvement in the very North Pacific, which I'm not really aware of, but March of Romain as we saw has a dramatic improvement. But again, the dynamics are slightly different than March of Romain. What I'm going to show briefly now is regional patterns of what is blocking results in, in terms of the meteorological response, temperature, precipitation, and the way I'm going to do that is, for this blocking metric, I'm going to construct a kind of composite picture where we're looking at the strongest to the right of this dotted line and the weakest to the left, 10th percentile, just CSM1 in this case, not using the mod and comparing that in Europe and Pacific to analysis. So this is what you get if you take the strongest 10 percent and weakest 10 percent in the Atlantic and Pacific, this is the geoprocentral height field that you get for observation. So it's pretty large scale, there is associated obviously for Atlantic, La Jolla is an Atlantic high, similarly for Atlantic high, West Pacific high, there's a low anomaly to the south and similar to the Pacific. So the question is, can we capture the response at these two, in these two regions? And so this is now a good example summertime. West Pacific here, if we look at this composite as I just showed on the left, what's the response to get in summer? And so it's warmer in summer, is what I expect over Europe, slightly colder over the Mediterranean and precipitation much drier in a model and then much wetter over the Mediterranean. So these are, if you can get the blocking, these are the kind of meteorological patterns you really want to catch obviously for extremes and impacts. And this compares pretty favorably to your interim, kind of surprising some of those. So it's kind of a nice example that, you know, regionally this is well captured. In winter time, not so much, which is surprising from the perspective that it's actually the signal is just stronger for the tail of the model that's greater than 90% percentile. Temperature distribution is pretty good, you know, regionally and along all the way over to Western Europe, but it really does capture the substream. But the precipitation is not as good as the Western Europe signal, way too strong. Another example of, you know, a poor kind of composite picture, this strongest blocking, the weakest blocking, it's just blocking. A good temperature signal for blocking events over the Northern Pacific, but over, in terms of precipitation, the model, and I don't know quite why, but the model seems to think that blocking is associated with a ton of precipitation signal over the Western Pacific. And again, that's troublesome, because if you do capture, for example, you know, the Pacific North American P&A pattern pretty well, then you're kind of in trouble because your precipitation signal on the west coast of the US might be around this, so that's something to watch out for. All right, so I finally got to teleconnection, sorry for taking me a while. So where does this all sit in terms of, you know, what I talked about, NGO and the connections through the Northern Hemisphere, and can we predict blocking? Can we predict the teleconnection pathways? And so, yeah, as I said before, the institute short-term forecast finds it very hard, so the question is, can we do anything from the perspective of blocking on these timescales? So this is kind of a nice graphic of, you know, MGO type circulation, convection in the Pacific, Indian Ocean, how that moves into the Central Pacific, the upper level diver, you know, this is the train of, you know, canonical ways that I'm sure you've been told about already during the colloquium, so essentially, you know, convection, heating, upper tropostrate divergence through a process, through a mechanism called Rosby wave source generation impacts circulation in the North Pacific, like I said, low wave number, Rosby waves, and these kind of shaded blue, orange, blue, kind of P&A type pattern. And then there is some, I'm never quite sure what these dotted lines are, impact into the North Atlantic. So, and this is describing in the North Atlantic, the North Atlantic constellation of which blocking is a component of that, so I'll go into that in a second. But the key thing, you know, the Rosby wave source is used on seasonal timescales, and I'll show you an example in a second that's used on, you know, subseasonal timescales. So Rosby wave source essentially says, what is the impact of the divergent circulation and the mean, both the anomalous and mean component of that, and, you know, the rotational circulation? And there are a number of terms that basically say, this is the source of the Rosby wave activity, you know, involves divergent, mean divergent, anomalous and mean vorticity times and anomalous divergence, advection of mean vorticity graded by anomalous divergent. When this is more understandable, because the divergent from the convection, you know, really impinges on the mineral gradient of vorticity on the jet. So these, particularly these two terms are the most important. So this is a really nice study that I came across. And basically, it's showing an example of can you connect the MGA activity to North Atlanta isolation that presumably blocking. So this is what happens here is there is a simulation of around 180 days, six months from October 1. And simulation is a cam. But what you do instead of letting 50 of these simulations run and seeing if you capture the MGA, you impose the heating, you know, all of these simulations are imposed the MGA heating. So this red and blue, essentially is the very integrated heating. And that's imposed from a, I'm not quite sure, exactly the year, particularly year. And so this is a succession of events that relates to this schematic to the right. And so these events propagate the black lines here. So other Rosby wave source. So soon as it gets to around 120 East, that impinges on the jet and the divergent flow as an impact on the circulation is generating Rosby waves. Further to the wet, to the east, these cyan type lines are the jet kinetic energy. So then not only are you forcing a P&A type pattern, you're impacting the jet's kinetic energy. And ultimately, you know, it's the connection between this and this that is that is a hard to understand and be hard to determine based on the fact that isn't, you know, the NAO is quite noisy, and you need 50 simulations to get any kind of signal. But ostensibly, you know, you can say, you know, once you generate the Rosby wave source at around 120 East and an impact with the jet, then later it's slightly more baroclinic when you look at these 50 simulations. And also during the kind of quiescent phase, oh, sorry, got that the wrong way around, during the active phase, then, you know, the blocking baroclinic is about even, but that's just saying the blocking is more than clontological. So that's this, these signals are very small, there's a lot of noise. And again, this is related, this is based on perfect heating from the MGO. So, but of course we know as IME said, there's a long way to go between initialized simulation and near perfect heat. You know, finally, there's a rich, sorry, can you wrap up in two minutes or so? Yeah, I think I'm just, yeah, I think that's it. Yeah, two slides. There's a further complication is that, again, that's, it's based on a series of success, succession of events to, you know, have to occur in terms of getting this teleconnection. So this is an example we've been working on when we look at the fives and the sixes are AMIC-5, AMIC-6 type simulations, so they all have the same SSTs. And this is, on the left here is the divergence anomaly related to the Roswell resource. So, clearly, you know, the analyses lie here at a 1.5 times 10 to the minus 6 divergence, and that gives you a Roswell resource, you know, on top. And some model occupied this same phase, you know, AMIC-6 and AMIC-5. And so we can see here, but there is a, there's a good relationship between Roswell resource and divergence, but of course, the divergence is not great. And, you know, I'll flip down to this right one here and say climatological divergence here is poor in CMIT-6, sorry, in CSM-2. So a number of these models are quite poor. And then, but they have a relationship that's pretty well defined, but then you say, what is the relationship between Roswell resources here, all four of them that I showed, and the P&A index. So that's the forcing of the P&A index from the divergent flow. And so, yeah, it's all over the place. There's a lot of work to be done. There are some, you know, the lieworthy analysis, but the forcing of the P&A, you know, which is critical in terms of thinking about the jet activity and then, you know, impacting, blocking downstream and also in the North Pacific maximum. It really is hot. It really is is not much skill at all in these models. And this is my last slide. And it's, you know, it's the only tropical slide I've shown, but I think it relates to some of the questions before, actually. Again, you know, that, that similar, that study of those 50 ensemble levels that gave rise to, you know, some small signal in blocking versus non-blocking, you know, they're predicated on, you know, getting good tropical heating as well. And this is just, I think, a nice example that shows, again, this is CAM-5 used in these initialized simulations for a period shown on the right here that has a number of MGOs during the period. And we can see that what, what are the biases at different periods in this, in the kind of lead time? So day one, day two, day 10, or for all the, for all the forecasts through this period, you know, it's, it's complicated. And like, as we talked about, there's a certain, you know, day one, there's a, there's this dry shock, if you like, the convection. Dry shock, the convection does not respond to the analyses, which is, you know, obviously, what we haven't talked about is, this is not, none of the STS are native, native analysis. They're all analyses that come from either, you know, Mary, Europe. So there is some initial shock, you know, in terms of thinking about, it doesn't like the analyses, but then it recovers by day two and overdoes it, it would seem. But then by day 10, this is now where the model biases are coming in. So, I mean, it has to be stressed and I'm sure it's distressed. Model biases, even before day 10 are really, really appropriate. If you, I feel I looked at the model bias field of these anomalies, it looked pretty similar. So yeah, I'll just leave this and yep, it's what it's, and, you know, the last two lines essentially are the key thing in terms of, you know, progress in terms of prediction are related to a number of successive events, successive connections along, along the teleconnection pathway, which is not, you know, which is not new, but I think tropical vegetation, divergence, jet interactions, they're all part of the barrier intended, getting good blocking prediction on these timescales. I'm losing my vice. So thanks. Thank you very much, Rich. Thank you. Yeah, thank you for introducing us to blocking and, and I wasn't sure if it was talked about. So it was great. No, we, it came out, but it wasn't.