09:57:45 We're going to switch gears now and move back to the atmosphere.
09:57:54 Here we go. Okay, so uh, Peter hands as well known to many of you. He's a professor in depth at the University of Cambridge where he leads the open atmosphere ocean group is made pioneering progress and understanding fundamental processes particularly
09:58:09 in atmospheric flows atmosphere ocean coupling dynamics and the dynamics of flows in the giant atmospheres and today he's going to talk about how important jet structure for atmospheric dynamics and transport.
09:58:21 Go ahead, Peter.
09:58:23 Okay, thank you. Bruce for the opportunity to talk.
09:58:41 So the first thing I thought actually what I explained what does my title mean, I've decided to kind of rephrase My title is what questions remain in scope for simple fluid dynamic models and what required detailed physics.
09:58:45 And so I'm going to talk about kind of atmospheric examples of the kind of dynamics and transport phenomena that's program is covering.
09:59:01 But I'll still often sort of what I, what we might regard as common ground okay so the you know this this is a picture like and I'm sure that many other people have shown us the two plane turbulence, the kind of canonical sideways layer forming system
09:59:20 in which one sees jets form as and manifest as staircase in potential dusty and one sees corresponding structure in tracer because the Jets actors transport barriers and these this picture on the right is histogram display of the, of the distribution
09:59:49 of potential testy and tracer you can see that the structure is somewhat different. In fact, the facility that the DP Susie being applied to the potential list here which is larger than that of the trace that the point was that, you know, there are some
09:59:57 situations where somehow the barrier effect on the tracer is not so limited by the sharpness of the Jets.
10:00:09 What can they said these sorts of simulations to multi level flows. So this is an example of a system which we considered quite a while ago now, but essentially one is considering a, a flow with vertical structure so this is in the world of quizzes just
10:00:25 traffic dynamics rotating stratified flow.
10:00:28 We've, we have, you know, this is the kind of latitude the y coordinates.
10:00:34 And this is showing the transport and mixing structure that you see in this flow, and this is a very connected and stable flow.
10:00:43 But as you know, the many, many, many levels in it.
10:00:47 And what one sees is that the flow naturally organizes itself into a kind of upper level and lower level transport regime.
10:00:55 The lower level transport regime.
10:01:05 This represents a broad strong mixing region, and the upper level transport regime. There is a. If you like a mixing barrier at the center of this region and surrounded by mixing regions the mixing barrier is, you know, the leads is part of the formation
10:01:14 of a jet so here you're seeing a flow which is it through its internal dynamics.
10:01:22 Determine the transition transition height, if you like in the transport regime, rather than the same. Along the same lines as if we look at a beat a plane to the turbulence about going to load Berkeley turbulence The, the distance between the Jets or
10:01:38 the barriers is determined by the internal dynamics here it's not only the few like the width of the barriers regions or the distances between the jet it's also the height of which this transition occurs that is determined by the internal dynamics.
10:01:57 So now let's talk a bit more about the real atmosphere.
10:02:01 So this is a kind of sort of much used way to display transport and mixing properties.
10:02:10 We're seeing a picture here which is extending and latitude from the South Pole the North Pole.
10:02:16 In Heights by about 30 kilometers going from the troposphere down here up into the stratosphere.
10:02:22 This is northern hemisphere winter, this is southern hemisphere winter.
10:02:28 The, the red, the red regions indicate whether it's strong horizontal mixing.
10:02:35 The, the blue and green colors indicate regions where there is a barrier effects if you like. And so there are some important barriers in this flow.
10:02:47 I mean one, which is well well known is the the barrier, the jet is, which is in the winter stratosphere. Okay, so this is the northern hemisphere polar vortex.
10:03:00 The mixing barrier associated to the edge of that jet in the southern hemisphere winter there is a stronger vortex and a stronger mixing barrier.
10:03:12 I'm actually what I'm going to do in this talk is focus more on the upper troposphere lower stratosphere region so this is that about 10 kilometers so this this black line is the notional boundary between the troposphere and the stratosphere.
10:03:28 So this you know this boundary characterized is differences in stratification week stratification strong relegation the stratosphere.
10:03:37 It also characterized as changes in chemical character.
10:03:41 You know, this is low ozone and moist air, and this is high Oh ozone and dry air.
10:03:51 And then, so I've marked on these, you know these locations in these pictures where there is there a jets and consequently or to corresponding me there also transport bears.
10:04:05 Um, so I suppose that one, you know one aspect of sort of idealize models is that one kind of starts off with the idea that in the atmosphere of the ocean.
10:04:15 Yeah, one has very high Reynolds numbers.
10:04:19 Why it's sort of inevitable that it wasn't a inevitable that I kind of lab experiment or a or a numerical experiment with somehow be too disappointed.
10:04:31 So, you know, it's one tends to think of these kind of simple flows, as being relevant to the atmosphere in the ocean.
10:04:39 In a limit to where dissipation becomes a small as possible.
10:04:43 Because in the atmosphere in the ocean, they're all really mixing processes.
10:04:49 And at some point they become important and of course if we take the example of ocean stratification you know then we recognize that small scale mixing ism is a leading order process and determining the stratification of the interior of the ocean.
10:05:04 If we think about the
10:05:07 you know the atmosphere here These are some examples of them water vapor observations and modeled distributions. This is in the upper troposphere about 12 or.
10:05:19 So kilometers.
10:05:21 You know, these are satellite observations showing the latitude on distribution of water vapor, and this is a, an average overseas and this is how much you'd hear.
10:05:33 This is now, nor nor the house for summer season is it winter, this is not have known her winter, summer, summer, there's interesting seasonal variation.
10:05:44 But I think the thing I just wanted to remark in here is that this, this is a nice investigation by partially via at all.
10:05:54 They're from the, the, you link, and they use a numerical model called clams, which is a kind of liberal engine based model it basically, it's particles plus recruiting and the recruiting is intended to represent mix it now you could, one could ask one
10:06:12 question how realistic the mixing is as represented by the recruiting but the model.
10:06:19 The model doesn't suffer from the, from the usual problem of, kind of grid based, and you're there in grid based models that mixing is inevitably too strong and they can essentially choosing the mixing in their model by changing the parameters defining
10:06:34 this regretting.
10:06:37 And one of the points that simulation is making actually is that there's some evidence that in order to reproduce these distributions.
10:06:47 You don't want to zero and mixing, I mean zero mixing is not
10:06:51 the right limit you need finite mixing. Okay, and so so that said there is no, this is a case for. There is real mixing and actually you need it to get new distribution right.
10:07:06 I mean these are the kind of notional values of this is a vertical diffuse 70 which has been a sort of equivalent vertical diversity to this recruiting.
10:07:19 And you know this, so this is representing and mixing which is taking place in the vertical associated with patches of turbulence.
10:07:34 You know, various authors have tried to estimate what this value is. I mean I'm at a camp who think that about 10 to the minus two meter squared per second is about right but I mean just as when considering ocean mixing and, you know, there's a lot of
10:07:46 of debate about how you go for example from small scale observations of small scale turbulence to an estimate of vertical the facility.
10:07:57 Okay, now let's move on to
10:08:01 weaken potential disagreements and forecast era so if you'd like. Now we're back in a world of them with we're thinking this is latitude.
10:08:09 What we're, we're visible it's being visualized here is the is the tropopause it's just, I mean it's that it's the contrast between stratosphere, and troposphere on a particular quasars onto the surface, its associated with a with a sharp potential just
10:08:28 a gradient which we it again in this program would recognize as being associated with it with a jet.
10:08:35 There's a group of reading, Sue gray and colleagues, who've been fighting again quite some convincing i think that that there's a significant sort of forecast era is being caused because the models can't sustain the potential just a great yet okay so
10:08:54 this is, this is if you like the analysis, this is what the model is has it says initial condition.
10:09:00 If you look at, you know, how the model is representing the TV distribution.
10:09:06 After a few days, you know you can see systematic differences and those differences, include a sort of systematic smoothing a systematic weakening of the gradients.
10:09:17 The This is the, the decrease yeah this is a lead time here of different forecasts and shows the, the weakening of the potential just a gradient with time and the model.
10:09:29 And this affects the evolution of other systems, I mean there's a couple of papers again by colleagues of Sue gray at some reading, looking at simple models where you look at the effects of weakening potential distributions on raspberry waves.
10:09:45 This is fairly simple dynamics and.
10:09:49 But if you like you, bye bye mate by going to the looking at these forecasts and then thinking about the weakening of the potential list of gradients, they reckon that you get, you know, something like a 400 kilometer phase era and your Raspberry propagation
10:10:03 over a period of a few days and so if you're thinking of weather system for example propagating across the Atlantic.
10:10:10 And you can see that these kind of errors. These this kind of phase there is going to make a difference to your forecast.
10:10:20 And, again, going back to work by Sue gray I mean, this is all about the forecasting of blocking which is one of these phenomena where the jet gets strongly displaced.
10:10:33 You know characterized in in Northern Europe by on, you know, hot warm weather in summer and cold, dry weather in winter. It's always been a challenge for models to forecast blocking.
10:10:46 This is inflammation the shading is kind of observed blocking frequency as a function of longitude and the. The colors are different forecasts models.
10:10:57 The, and this is now working through in terms of the lead time of the forecast okay so the message here is that the you know the modeled distribution of blocking decreases the blocking becomes less frequent as you as you go on in the forecast right the
10:11:15 forecast, the deteriorating.
10:11:17 And one of the pieces of evidence that Martinez over audit, and they tell us here is the fact that the metal is changed their their dynamics in this period.
10:11:30 And this is their dynamics from the two years in the period being studied and this is that using a different model the dynamics.
10:11:38 One of the effects of this different model dynamics is to
10:11:42 maintain potential just to gradients.
10:11:47 And you can see the What seems to be the effect of that in the improved performance of the metal is model.
10:12:06 the middle is had the old dynamics and their model. And this is when they have this new dynamics and they're the red line is the metal base that their forecasts seem to stay more skillful for longer.
10:12:21 Okay now, in some sense, the kind of weather forecasts problem you, you started with initial conditions you integrate with some limited time is. It's a, It's a sort of somewhat it's a simple problem.
10:12:35 You could you can track the you know the deterioration the forecast, you can identify phenomena that are leading to the deterioration in some sense the climate problem where you're thinking about the simulation of the system over a much longer time scales
10:12:53 is is more difficult because you know you're not simply diverging from an initial condition, you're the you know the climates that the statistical distribution of dynamic variables is being determined by internal balances between, you know, different
10:13:09 processes between a decent turbulence and.
10:13:13 And the one handed mean, mean flow on the other.
10:13:15 And so that's understanding why models, make errors in that is, you know seems to me to be more challenging.
10:13:24 There's a phenomena which is causing a lot of interest in the seasonal forecasting and longer term forecasting museum called a signal to noise paradox, which basically comes down to it, it's part of a question of is the internal variability of climate
10:13:40 models to large.
10:13:43 Bruce How am I doing for time.
10:13:53 Okay, good. I think you have eight minutes sorry Peter I was muted. Got you.
10:14:01 Another 10 minutes.
10:14:01 Another 10 minutes. I am so I have time to explain this picture, I think it's very interesting picture this is from Adam scape his head of seasonal longer term for skin forecasts in the mid August.
10:14:13 So this is a set of seasonal forecasts, you're starting the beginning of each engine your forecasting the winter, and their forecasts good particular measure of the circulation, the North Atlantic isolation.
10:14:25 And the, the black line is the is the observation, so you know how it is varied over two decades,
10:14:36 the, the, the orange line is, is the forecast if you like so. So, what this orange line seems to say is that, you know, if we start at the beginning of each winter and and forecast and then we we actually can predict to the observer ability there isn't
10:14:54 a prediction being made from 1992.
10:15:02 Early in the winter in the autumn or something.
10:15:05 But then when you look more carefully at this picture you realize that mean the this orange line has been generated as the mean of an ensemble of focus right and the ensemble or shown by the dots.
10:15:18 And, you know, by any sort of rational explanation interpreted what's going on.
10:15:22 I mean model world would be, you know, choosing one dot from each winter at random. Right.
10:15:32 And of course, that would be a much more variable pattern than what you're actually observing so there's a mystery about why the, the atmosphere if you like in a given winter decides to follow the prediction of an ensemble of focus.
10:15:54 Say this comes down to a question about is the internal variability of the clock flow models to large it has other implications as well.
10:16:02 And this is all potentially to do with the dynamics of the variability of the jet, and of the potential just the interface of the other balls.
10:16:20 There is a sort of very slight hint that the problem may reduce as you increase resolution, there isn't really time to explain this picture but I mean we're talking here about going to resolutions, it was sort of order 10 kilometers or something you know
10:16:26 this is pretty challenging pretty high
10:16:31 global model and certainly out of reach for most climate simulations.
10:16:39 So the question is, you know what, what can one do about that how can they might understand it I mean one question is well if we understand the forecast era problem does that tell us about the climate era problem or, you know, possibly yes possibly no
10:16:59 I mean the climate every problem is all about this two way interaction or it's an auto interaction between many different processes.
10:16:59 I suppose one of my one reason why I like this kind of beat up lighting service model is that's the sort of simplest possible model.
10:17:07 It deserves to be in textbooks on this basis, where we capture the interaction between turbulence and waves, and the mean flow.
10:17:18 And therefore, potentially, allows one to, to address some of the problems with, with climate simulation of this type. I mean, Laura.
10:17:29 Pat diamond asked. You mentioned Laura's talk earlier I mean, Laura. Koch gave a talk on variability, couple of days ago, and now why is variability interesting well I mean of course their ability is interesting and fun on its own right, but I mean, I
10:17:44 find it particularly interesting because the fluctuation dissipation theorem tells us that there's an intimate link between variability and forced response, and that the you know the, they're responsible for the system can be expressed as the kind of
10:18:10 correlation time in the system, times the four sec This is very very sort of naive and simplistic actually you've got a, you've got a multi dimensional system. So, so if two models have different variability you expect them to have different climate response
10:18:21 So that's about time for me to stop.
10:18:25 It seems to me that the dynamics of PV staircases is, it's kind of you know it's there in several different aspects of atmospheric transported endemics.
10:18:46 It's important for dress moderator to be active species such as such as water vapor, I'd want the I didn't say was that if you make arrows in the water vapor in the upper troposphere chosen region then you make, you know significant heating arrows and
10:18:48 that means that our circulation or is.
10:18:58 It's also really relevant to to shorten meter means weather prediction and to these problems of long season or longer term prediction. And so the question if you like is. Well, where is there a role for simplified fluid dynamic models.
10:19:04 And where is it best to simply sit back and do something else and wait until complex models can be run at higher resolution.
10:19:11 And I'll stop there.
10:19:13 Very good. Thank you so much, Peter. I want to start by asking a question you talked about how the Met Office, changed the dynamics of the model to reduce the week and again PV.
10:19:23 I wonder if you can elaborate on that. Do you know what dynamics have changed in particular. Well I shouldn't say they changed the dynamics to reduce the variation of PV right.
10:19:32 They, they changed the dynamics in it as part of a longer term program
10:19:40 to turn it to, I guess, to improve various, various aspects, perhaps think things like a mad faction of dynamical and rages of actually traces or whatever, right so I think your week, and that there is, there is a their current system is called end game
10:19:56 I think you're better off.
10:19:58 You know, if you going to read a paper on end game rather than me, trying to describe what it is but actually what I'm one of the points that's made in the paper actually is that this improved dynamics doesn't require a higher resolution so it's the kind
10:20:12 of, you know, it's, it appears that if you formulate your dynamics better.
10:20:18 You know, then, then you can actually get a significant improvement, even at the same resolution.
10:20:25 That's kind of where I was coming up with the question it's not, it wasn't a resolution issue or just changing viscosity, but it was something more about the change in the way you formulate represent the dynamic equations.
10:20:36 Yeah, in this sort of in the world of a fight a finite grid etc.
10:20:41 Right. We have a question for part.
10:20:46 I think Rick Sammon was first and then and then the okay yeah, Rick Rick put his hand down, Rick Do you want to ask you a question, actually my question was the same as yours Bruce, and so thank you guys think of like, okay, but.
10:21:00 Thank you think alike, okay but. All right, well I'm, I'm sorry but me doth protest a little bit about the statement about the fluctuation dissipation theorem which is of course, you know, and near thermal equilibrium creation, right, and you're in it,
10:21:16 you're in a situation here which is strongly nonlinear.
10:21:21 And, in fact, we've come across this issue in recent work.
10:21:27 And really I mean, I think it's better what's going on is better thought of a balance of in the jargon coherent and incoherent mode coupling.
10:21:38 And the dissipation is really oversaturation of the collective mode, you know it's not it's not quite so simple as the, what you find in textbooks for the for the fluctuation dissipation theorem.
10:21:53 Can I come back to you on that right i mean so that actually, you know my view is that, I mean, one can write down to what I would call it a more general fluctuation dissipation during which doesn't require gal sanity.
10:22:08 And, you know, it's essentially a statement that the response to the fourth thing is, is determined by the fluctuations.
10:22:22 So whilst it's true that the textbook formulation of the fluctuation dissipation theorem is generally written for systems in terms of equilibrium, and is essentially, assuming a Gaussian probability distribution and face space or whatever, right, one
10:22:38 can actually generalize it.
10:22:42 There were questions about whether the generalization is practical, in the sense that, of course, if you want to compute the response you have to compute more and do more more delicate measures of the variability, but I would say, I like it is not the
10:22:59 case that the fluctuation dissipation during in a general sense requires them are dynamic equilibrium.
10:23:08 But I mean, I really, I think it's misleading to cast it in the, in that that framework because what's really going on is a balance between different parts of the interest scale coupling.
10:23:26 Right. And it's, it's much more complicated than the simple fluctuation dissipation theorem looks ultimately you have a balance of course in a steady state so maybe you can manipulate any balance into a fluctuation dissipation theorem, but really you've
10:23:48 parted company with the, The, the one we usually appeal too long ago and systems like this.
10:23:56 Yes, I'm certainly appealing to a significant generalization of the standard fluctuation is a patient term right and you might prefer to call it a linear response theorem for dynamical systems or something like that.
10:24:12 Right.
10:24:12 Let's take a question from Boris.
10:24:15 Yes, it's common rather than equation.
10:24:19 We've published a paper recently about the physical nature of the master engaged spectrum. And it comes with a different interpretation of the spectrum.
10:24:31 This is on large scale.
10:24:34 The key to minus the spectrum was determined by this parameter
10:24:41 transfer. And if this is a key ism, it takes everything much more hardware physically what is parameters is made is the reason why I did it is not flattering because those parameters.
10:25:04 so it was no research on it.
10:25:19 choice I think they possibly didn't hear a part of that.
10:25:27 You can you can you hear me now. Yeah.
10:25:30 I'm saying. In, we have a new analytical theory of mastermind over a teaching tool, which predicts the amplitude of the mastermind gay spectrum scales is determined by the square of the Coriolis parameter MK to minus three.
10:25:49 And if, indeed, the large scale dynamics of salt hardwired to the Coriolis parameter, it will reduce its ability.
10:26:00 So it will reduce the noise on the system
10:26:05 would be interesting to hit dare to touch you can send me a link, link to the paper.
10:26:09 Yes, of course.
10:26:12 Okay, let's let's take a question from Oliver, Peter.
10:26:16 Could you go back to your slide was a signal to noise ratio, I had a question about that.
10:26:24 Yeah, that one. So you pointed out that if you sort of just pick one of the orange bubbles at random, you would get a trajectory that would look a lot less accurate than the kind of ensemble average there.
10:26:36 But isn't it, isn't it, I mean I'm just thinking about that isn't it the case that the atmosphere would not pick a rapper orange ball at random, it would be picked one was the probability that's proportional to the probability density at that location
10:26:51 right so you see a lot of these points are space closely together, and the atmosphere is much more likely to pick one of those, simply by the density effects, I don't think, averaging over all means sort of have to endow each member of the ensemble was
10:27:06 somehow a measure of how possibilities right i mean kind of how likely, it'll actually is some kind of the system is going to take that state. Does that make sense, or if you just look at it you sort of don't look at the balls just to kind of the width
10:27:18 the boards was try to imagine the PDF that that covers them, of which there are a sample, then that doesn't look so bad but has a different clustering near the most of these points near the mean.
10:27:32 So I've kind of feel that.
10:27:34 Maybe it's not true that the atmosphere would pick any of those realizations at random. You're right, you're drawn to your drawing from a probability distribution.
10:27:44 And so, it's, it's true that it's not quite fair to say that you're picking one of those points at random, but I guess that it's not.
10:27:52 It's also not obvious that you, you would know that the most likely ball to pick is the one which is that the mean of the ones you can.
10:28:04 But the most likely to pick is the one that you, which is the mean of the values you can see that right so no it's unclear whether mode of the distribution would be better somehow, thinking about the distribution that, which these balls as samples.
10:28:20 Might be quantitatively addressed with how big the paradox is, I'm not understand what you're saying, I'm just wondering exactly how a quantitative measure would work.
10:28:34 Thank you.
10:28:37 That might be a way of resolving the paradox.
10:28:39 You heard it here first.
10:28:43 All right, let's thank Peter again for his talk and Peter can ask you to stop sharing your screen.