Loading summary
Transcript131 lines
- [00:00]
A
I'm your host, Ed Porter. Welcome back to Transmission. There's an El Nino building in the Pacific that some models say could be the biggest in a century. That's warm water off the coast of South America. So why does it matter to a power trader in Europe? Because weather is the input the entire energy system runs on. And this episode follows the chain from sea surface temperatures all the way to power prices. Dunkel flouters, French nuclear shutting down when the rivers run too hot. AI models versus human forecasters. My guests are Emma Patmore and Matt Dobson, meteorologists from Met Desk, the people who turn the weather forecast into something an energy trading desk can actually act on. Want to know what the next dunkelflauter could do to German power prices? Ask Co Moto Energy's AI analyst link. Link below. And if you're not already, make sure to give us a rating and follow wherever you listen. Let's jump in. Hello, both. Welcome to Transmission.
- [01:03]
B
Thank you for having us.
- [01:04]
C
Yeah, it's great to be here, Ed.
- [01:05]
A
Our pleasure to have you on. I've been wanting to do a weather episode for ages and now is finally the chance. So, Matt, I'm going to come to you first. What does everyone get wrong about weather forecasting?
- [01:17]
C
Okay, so weather forecasting is a complex science. There's lots of variables at play. The atmosphere is a chaotic system, but I think it's only in terms of the energy industry. One of the things I feel is that they assume that forecast skill or predictability is static. So a day ahead forecast will always have a certain element of skill. Trying to predict, say a month, three months ahead will have a certain level of skill. So it's that kind of sort of myth that we're trying to break down when we're consulting with clients, that the variability in skill is quite large. Sometimes predicting a day ahead is easy, other times it can be very, very challenging. And likewise a seasonal forecast. Sometimes there are windows of opportunity and that's what we try and get into. Energy traders, energy analysts, heads that there can be these opportunities to predict maybe four months ahead quite skillfully, but not all the time.
- [02:06]
A
Okay, yeah, well, that's. So a really obvious question is when you're talking about skill, do you mean the success of the thing that you're doing? So let's say the day before you say the wind speed is going to be 15 meters per second and then we get into that day like what is a good skill? What is a good outcome? Is it sort of accuracy in sort of 10% away from that or like, do you feel like you could predict it down to something more narrow?
- [02:30]
C
That's right. I mean, obviously wind is a little bit more challenging in temperature to predict. So for example, a day ahead, wind forecast, let's say 80% of the time, the arbitrary figure, you should be within maybe 10, 15% quite commonly. But now and again you can get weather patterns that can introduce sort of uncertainty and chaos into the pattern. For example, an area of low pressure, that's bad weather, that's cloud, that's wind, that's rain. If that tracks in a slightly different position, you can end up having maybe 30 or 40% less or more wind than if it tracks maybe slightly. We're talking like miles here, 50 mile north, the low pressure could bring a lot more wind. 50 miles south, it could bring a lot less wind. So these are the events we watch for to give clients an area of idea of risk and likewise temperature. You know, trying to forecast that maybe a month ahead. Now and again there are these windows of opportunity. But wind is harder to predict temperature.
- [03:19]
A
So your list of what's easy to predict, what's hard to predict temperature is sort of relatively easy. And then kind of like you're moving towards wind is slightly harder. Where do things like solar fit into that?
- [03:28]
C
Yeah, I think relatively is a good word. I mean there's always opportunities to get, you know, for the weather to be wrong, or rather our forecast to be wrong and the weather to do something different. But temperatures are more sort of evenly spread parameter, if you like, compared to wind. And wind can be affected by lots of different things like topography and for example, individual weather systems. Okay, solar, that can be easy sometimes, but it is affected by things like snowfall. You're thinking about solar generation, you're thinking about solar panels. How can the solar panels be messed around with by other things? So for example, dust from the Sahara, believe it or not, can get blown up towards Europe.
- [04:08]
A
We had that a few years ago. I remember seeing it sort of on the cars like this, kind of like this clay, like dust that was on everything.
- [04:14]
C
That's right. So you go and get your car washed on a Friday, you get a Saharan dust about on a Friday night and you're very sad on a Saturday morning.
- [04:20]
A
But do you have a forecast of Saharan dust?
- [04:22]
C
We do on our Met desk. Have now taken the cams, the Copernicus aerosol monitoring and we have a five day chart looking at the dust. So dust is one thing. Snowfall, you can get a lot of snow on the panels. That sometimes doesn't melt. That can have an effect. So you've got all these other factors that are not cloud related as well.
- [04:40]
A
Can I ask the Copernicus aerosol monitoring. I hope they got that right.
- [04:44]
C
Yeah.
- [04:44]
A
Is that, that is a satellite looking down or is that a difference?
- [04:49]
C
That's a weather model that specializes in fine particles of dust in the atmosphere. So you're looking at something very specialized and obviously it's a three dimensional thing, a plume of dust. It's not just a thin layer, it's like all different levels of the atmosphere.
- [05:04]
A
This is, this, this is crazy. I've got real concerns this episode might go to like two and a half hours as I ask loads of really stupid questions about things that I think people in the weather, the weather space know really well. I was going to ask one more question before I come to Emma on the Dunkelflauter in a second, but you mentioned that sort of like across like a 50 kilometer space, you could get really different results. Is that sort of, is that the level of detail you can go into? So, you know, you could be sort of 50 kilometres away from someone and you could be seeing a totally different day's worth of weather.
- [05:33]
C
To be honest, it can be even less than that. I mean, think for example, you're out on a bike ride or a walk and there's a thunderstorm approaching. Some places are going to absolutely hammered with heavy rain and lightning and hail and then literally a couple of miles away or even less, you may miss it. So the weather operates on so many different scales. Frost, sometimes you can get frost in a valley, you don't get frost a couple of miles away on a hill. So there are those kind of very narrow scales that make even intraday forecasting challenging. But on the bigger scale, a lot of the time the weather doesn't change that much over 50, 100, 100 miles. I'm going to look out the window and say, well, today, especially in the southeast of England, many of us are getting a very similar weather on scales.
- [06:13]
A
We were, we were like, we're doing a podcast about whether we had to look out the window at some point. I'm glad you did it first. So, okay. And the thing that people get wrong, right, that you've talked about is that skill element. So sometimes you can look, say to the weather tomorrow and you can have a really high degree of certainty on it, but sometimes you could say, oh, actually I really don't have that confidence.
- [06:32]
C
I think that's right. Yeah. I think that's where the expertise comes in. That's why energy clients speak to people like ourselves because we're monitoring that kind of threat of the skill changing.
- [06:42]
A
Okay, okay, really interesting. Emma, I'm going to come to you with a slightly more energy specific question. So the thing that the Northern hemisphere energy systems worry about is this thing people call a Dunkelflouter, which is a German word that we borrowed and it essentially means that it's cold period. So demand is high, wind is low and there's obviously it's winter, so there's not much sun. It's quite hard for a renewable system to deal with that. So I just wanted to understand from you, so like what, what's the driving force behind that? What's actually happening from a weather sense to give us those cond.
- [07:13]
B
Yeah, I mean within a system it can quite often be high pressure pulling in and so you end up with those lower wind periods with that cooler air pulling in from the north, so from Scandinavia for example. And so you end up with that combination. And they can really vary as to how long they last for. So I pulled the data yesterday. We had a big spreadsheet put together looking at German wind, particularly there, and we were having a look, we took out the capacity factor of it. So we were looking at sort of megawatts per gigawatt. So that that didn't impact the numbers. And over the last five years, most of the events they were only lasting like two, three days. So they're not sort of longer lasting things. You get your low pressure coming in, freshening things up, for example, shifting out that fog. But there are examples where you do have that longer. So in 2024, November particularly, we ended up with a period of nine days where that just sort of sat over statically.
- [08:05]
A
And so that was 2024. We had nine days.
- [08:08]
B
Yes.
- [08:09]
A
Okay. Because yeah, I think like rule of thumb I had in my head was sort of a 14 day period every five years. But that seems like we're in a similar ballpark to each other there. So. But you said nine days. Was that for Germany or was that for the uk?
- [08:23]
B
That's Germany.
- [08:24]
A
Okay.
- [08:24]
B
So Germany is quite often where people tend to be interested in. So I focus the research towards that. It's a big market. They have most of the capacity in terms of wind and solar. They've got quite a lot. They've got about 100 gig now. So it's quite a powerful source of their sort of grid.
- [08:41]
A
And that is sort of statistically what's happening. You mentioned from a weather perspective, and you have to forgive me Because I'm not a huge weather expert, so I'm going to have to sort of ask you around this. But you said you had high pressure coming in. And what does that do? Does that create almost like a dome that keeps out change in the weather system? How does that sort of prevent the wind from coming through? Right. Does that make sense as a question?
- [09:07]
B
Yeah. So when you have this high pressure system sort of pulling in and sitting over the entire sort of northern side of Europe, if it's quite strong, it can be quite static. So any low pressures could, for example, slip underneath it, slide over the top. And so you're not going to get that sort of pressure gradient. So for that wind to come in, you want a low pressure system to be pushing against that high. And so you've got the gradient there to bring in. That's where you get your brazier scenarios. So when you're under this sort of high pressure system just sat overhead, you've got a really slack flow and so you're just not seeing that wind. And if you've not got the wind, you've got your sort of cooler conditions overnight. You start seeing that fog settling in in the. You've not got the wind to clear it out, and so you end up seeing it sat for longer.
- [09:46]
A
Okay, you're going to have to unpack this for me because I was looking at like a weather map, almost like in a 2D sense, but immediately I see that's wrong. And it's kind of a 3D piece and you've got sort of pockets of pressure and high wind going in the middle and low going over or above. How does. So just just to make sure I really understand it. So, so like to get a Dunkelflouter, what is actually happening? So I think I followed you up until this high pressure block effectively comes in and sits above a country. And then that is so strong, to sort of borrow a phrase, that it's kind of keeping out other weather systems or other sort of pressure systems for coming in. Is that right?
- [10:29]
B
Yeah. So imagine you had a map of Europe on your table and say you have your high pressure sat over, say, Scandinavia, for example. It's pulled around here. And so any low pressure system which is coming up against it, it could imagine it was completely flat. You could see it dropping down into Spain, going around it, for example, or
- [10:45]
A
it could slip over the top, pushed around the other way.
- [10:48]
B
Yeah. And so you need that sort of scandi either to pull northward to start allowing some slow pressure to come underneath it, for example. I suppose underneath is on a flat sense. So I mean sort of coming up more so towards like Switzerland for example.
- [10:59]
A
I'm with you.
- [11:00]
B
Starting to move northwards rather than.
- [11:02]
A
And why does it move? Like couldn't high pressure, couldn't a high pressure blob, that's not a technical term. Couldn't it just stay there forever?
- [11:09]
B
I mean you have different teleconnections at play. So different things which are happening elsewhere which drive the pattern. So for example, one that's cropping up lots in the news at the moment is the El Nino Southern Oscillation, which is a pattern in terms of sea surface temperatures that can drive a weather pattern. There's something called the Madden Julian Oscillation which is about precip and rainfall over and towards different regions of the world. And that again moves over can start encouraging more low pressure to push through. So there's many drivers which could be happening in a complete other side of the globe and influencing sort of what we're seeing. And so that's one of the things we look at a lot is what's going on elsewhere. Could that be driving things? Could that help change our regimes?
- [11:50]
A
Okay, so we shouldn't think about Europe in like a narrow. It's just European weather. We should be thinking about some of those global phenomenons as well.
- [11:58]
C
Yeah, I mean a good example is a typhoon or a hurricane. It's like a pulse of energy in the atmosphere if you like, and that can shake up the whole weather pattern. So you said could it last forever, the high pressure? Well, up until a point where there's a nudge from a different part of the climate system that could come from the tropics, it could come from America and that will shake up the pattern again. So there's normally a finite time, I mean sometimes three weeks, we get a heat wave that lasts for three weeks. Dunkoflatter maybe lasts two weeks. But there's normally something that then changes everything.
- [12:24]
A
So something else changes somewhere else in the world, gives it a nudge and it's not like it's a stable equilibrium. So it kind of keeps on coming back to the same place. It will just move on to another part of the globe. And you mentioned El Nino. So maybe let's do it, let's jump to an El Nino. I've been reading the news. Some very good news outlets cover it seriously. Some cover it not so seriously. I mean, how much actual science is behind the sort of long range El Nino forecast.
- [12:51]
C
I take this one.
- [12:52]
B
You can go.
- [12:53]
C
So I'm quite excited about this one, because it could be the biggest El Nino we've had in maybe a century. I mean, the reliable data goes back to 1950. How were we able to make that statement already? Well, El Nino starts under the water in the tropical Pacific. So imagine you're in South America. You go to the west of South America, you've got the big Pacific there. It's the tropics. And El Nino is when it becomes very, very warm between Australia and South America, effectively, when the sea surface warms up, there's a lot of energy loading up at the moment to make this maybe the biggest one we've seen since 2015. That was the last really big one. So first big El Nino in over 10 years. But potentially, some models are saying this could be even stronger than that one.
- [13:33]
A
Okay, and that means lots of energy in the water.
- [13:36]
C
Yep. Lots of warm water.
- [13:37]
A
What does that mean for Europe?
- [13:38]
C
So it flips the global circulation pattern. Normally, the water off South America is relatively cold. The reason it's called El Nino is that Spanish for the boy child. And it's basically a Christmas phenomenon. It comes in November, December time. And it means that all the fishermen that used to try and catch fish off South America suddenly had no fish. Fish like cold water, they don't like warm water, so they all disappeared. So it's a historical thing that was noticed, but it has a massive effect on the climate system. So, for example, it has a big effect on America and that then has a diluted effect on Europe. So I would say El Nino has some effect, particularly when it's very strong, but it's not the only thing that affects European weather. So we're kind of looking at it, not in isolation, but we're looking at it along with other.
- [14:19]
A
And what would we expect to see just from, like, we're going from solar, wind, snowfall, rain, like, how does an El Nino change what we see?
- [14:27]
C
To keep it simple, Emma might have other opinions, but to keep it simple for me is the autumn is when it has the biggest effect on Europe, the most reliable effect, and it tends to mean we have a warmer than normal autumn in terms of wind. You can get a heightened chance of winter in the UK as you go through November and December, but sometimes it can bring quite a calm period in, say, in mid autumn. And Scandinavia can go quite chilly. So the later you get into autumn, the earlier you get into winter, the more chance of wind events, mild conditions where maybe demand is lower for heating, but you're getting some supply. But across the Alps, it can be a nightmare for the ski season. So there's people going skiing for Christmas may suffer.
- [15:04]
A
So if you're taking notes, then skiing for the winter of 2026 could be a bit dodgy, certainly before Christmas. Yeah. Get a higher altitude chalet if you're booking. Okay, but this is kind of. This goes back to your first statement, right, which is that you've described El Nino in some fantastic detail with the historical references. But if I'm maybe, Emma, I come to you with this, like, if I'm on a trading desk and I'm trying to work out, like, what do I do from a power trading perspective with the knowledge of this El Nino coming. You mentioned that perhaps it could be slightly warmer. Perhaps there could be wind hitting Scandinavia as well. It might come to sort of southern Europe. It might come to Scandi. How do I. What do I do with that information? Like, is there a trade I could do? Like, how do I get better at my job from doing that?
- [15:54]
B
Well, we talk to a lot of people, obviously. And one of the things I really like to do is try and find out what information people want and just have normal conversations with them. And so different clients will use it in a completely different way. One thing that we do a lot in our forecast is looking at sort of a most likely scenario. So that's sort of what Matt's discussed there in the El Nino sense. What is the most likely scenario? You'll see. But there is also the alternative, which may be driven to something else. And so we have to communicate that risk. And they'll sort of price in and say, okay, well, this is our most likely scenari, Mari. This is probably what the market's seeing, but what if that's wrong? Where can we sort of make something from that? And so they'll look at sort of what we're saying. How do we think the models could move? Okay, well, we'll price in based on what the model's showing now. How do you think things could change over the next couple of months? And so can we get ahead of that model, move, react now and obviously then make the money when people panic at a later date?
- [16:47]
A
So they're trading, say, gas and power. And Matt's just said, right, biggest one since 2015. Biggest El Nino since 2015. And let's assume they're not exposed to Fisherman's Cash in the Pacific.
- [17:01]
B
Right.
- [17:02]
A
But they would think, oh, well, do I need to buy as much gas this year as I would have done in previous years? Because the temperature is likely to be slightly higher because the El Nino comes through in early winter so they should be less concerned about gas reserves. Is that the kind of that logic?
- [17:19]
B
But then you have to think beyond that as well. And so it's thinking, okay, in that time. Yes, that's what the temperature is, that's what your demand is. Okay, you might have some more low pressure in the northwest, so that pressure gradient will be there so you have a bit more wind. However, if you then say, okay, the Alps is seeing less snow than usual in the snow season, okay, when you get to melt season, that's not getting in your reservoirs so your hydro generation might be less.
- [17:40]
A
So think about the snow.
- [17:41]
B
You might start seeing your river levels coming down. And so maybe next summer you end up in more of a situation where your rivers, river levels are a bit lower. And so then could you be looking more at river temp? So it's not just the. In the now, it's right then that's what it's impacting. You have to think, okay, where could that then leave you going forward as well.
- [17:57]
A
Okay, so they're going short Q4 gas and then they're going long and you can start picking 20, 27.
- [18:04]
B
That's what Matt was saying earlier.
- [18:06]
C
Okay. Because you end up, Emma's already mentioned high pressure and high pressure is basically where the air is sinking and you can end up with a big mild high pressure sat across the Alps through quite a lot of the Q4, particularly November and December, which is dry and mild, which is something we shouldn't really see. You know, high pressure should bring chilly weather, but we're getting warmer and warmer air masses coming in. In the autumns these days we're getting these calm, mild patterns which can sometimes lead to dunkle flatter. But yeah, I'd be thinking temperature is probably the biggest. If I was forecasting and trying to help a trader, temperature is where I'd be trying to guide them. Mild autumn. But watch out for wind because we're not sure where that boundary is going to sit between the low wind and the high wind. That's too early to say.
- [18:46]
A
Okay. And this is, this is kind of, this leads me into sort of a question I really wanted to ask which is that you're meteorologists, right? So this is what you do day in, day out. I also know that companies have access to ECMWF data and I'm going to ask you what that means in a second. But that feels like couldn't, couldn't, you know, in this day of sort of AI, couldn't they just sort of like run that data through that and do they actually need someone who can give them context on top? Like, how do you fit into this to this world?
- [19:17]
B
Yeah, I guess technically, yes, you could just download it. It's the European center for Medium Weather Forecasting.
- [19:23]
A
Okay.
- [19:25]
B
But there is just so much available. I mean that's not the only model. You've also got the Global Forecasting Center, GFS uk, Met Office have a model. But even if you just took the EC data as a baseline, you've got 151 members. If you look at the different models they have, they've got medium range, they've got a longer range, you then got the season. There's just so much data available and making sense of that, every single member isn't going to show the same thing. You've got the uncertainty in there. And so when you talk to a forecaster, we can say, okay, this is what the pattern showing. We can link it back to how we think the model could evolve. We could say, okay, we think that low pressure is going to tighten up and say, yes, the plume has below normal wind at the moment on average. But actually you're probably going to get a brief peak as that system moves through which there's uncertainty on the timing. So the ensemble mean just isn't getting at. Okay, so it's that sort of detail that you can then start to add in there.
- [20:18]
A
And you mentioned an ensemble. What's an ensemble?
- [20:21]
B
So the run is sort of set off at time zero and you get the spread. So that's what the ensemble is. So it has say on the medium range model has 50 members. Each one will likely show a different sort of route basically of how you can move. So it's forced in the computation.
- [20:37]
A
Okay, when you say it's got 50
- [20:39]
B
members, what do you mean 50 runs? So it all starts at the same time, but you sort of run it 50 times, you end up with 50 lines of temperature, 50 lines of.
- [20:48]
A
Okay, so like an ensemble is like bringing together those, those runs into, like into one package that a trading house could say look at. And, and why are there like, why are there multiple versions of this? Because it feels like you said there were three groups that are, that could provide you with these ensembles. Like A, why are there so many of them? But, but also B, like do they match up? Do they all kind of say the same thing?
- [21:10]
B
Very rarely.
- [21:10]
A
Very rarely. Okay, well then maybe that's a good reason to have three.
- [21:13]
B
You see consensus. You can start being a bit happier, I guess with the general story. But it's that idea of communicating chaos of the atmosphere. I mean, the weather itself can change quickly. Everyone knows. I mean, if you look on your weather app, it quite often flips and everyone gets annoyed. I mean, it's trying to communicate that. So you're looking at all this data that's available and trying to make sense of a chaotic pattern. And so you need all that spread, you need to look at those different centers. Some do better at certain things. So for example, if you took an AI versus a sort of traditional model, sometimes the traditional can do a bit better on very small features just because the grid spacing is tighter and the resolution is higher. But if you want a higher level story at say the 10 to 15 day range, quite often it's the AI that is quite often in our experience, a little bit ahead.
- [21:58]
C
Yeah, I think it's worth clarifying here, Ed, about the AI models because meteorology has had a huge change in the last just few years with, you know, AI modeling alongside the numerical modeling. So numerical modeling is solving equations of the atmosphere, lots of equations which are producing a forecast. Whereas AI is looking at what's happened in the past using complex neural networks to piece together what's happened in the past to make a forecast. And as Emma was just saying, you know, there's advantages and disadvantages both methods, but having a look at lots of models, I think that's why you would employ the meteorologist and employ a forecast center like MetDesk, because you'd be pulling all this data into one place, analyzing it and then making it easy for the trader to digest that all in one go rather than doing it themselves. Having that expertise I think is key. And also developing products like powergen. You can buy all this data from ecmwf, but you're not getting the direct wind power forecast coming out. You're not getting what's my wind farm going to be producing next week. That's something that is needed to be done with people who've got the expertise.
- [23:00]
A
And I'm kind of fascinated in this topic that AI is coming into weather as well as the more deterministic models or the kind of equation solving the chaos in the sky, as Emma would say.
- [23:08]
B
So
- [23:11]
A
are things getting better? Is there a way of saying that? Because in the long run, if I think back, say 15 or 20 years, you know that weather forecasts, they used to be bad or worse and now they are better. And maybe there is a stat that
- [23:27]
C
40 years ago, a three day forecast skill, we're now the seven day forecast skill has the same predictability as a three day forecast 40 years ago.
- [23:36]
A
So we're as good at forecasting the seven day horizon as we were 40 years ago at forecasting the three day horizon. Okay, and is that sort of ramping up as AI then takes over parts of the modeling and are we seeing sort of increased accuracy?
- [23:50]
C
I think we are certainly in the important 7 to 15 day range and particularly the 10 to 20 day range. That kind of just on the boundary between where a lot of web forecasts tend to drop off in skill. Beyond about 10 to 15 days, the cycle of weather systems tends to become more chaotic and less predictable. The AI is almost trying to push the boundaries of that medium range forecast to beyond what we've ever had. I think me and Emma have still got a job. I think there's still an opportunity for human forecaster to assess the data and still come up with a valid view. And that's what we're doing day to day.
- [24:22]
A
I think it's just translating it as well. It's being able to say, okay, there are these models and if you get consensus in the models then it might work this way. If they're very far apart and it's this type of weather condition, then I'm just going to say it's sort of like risk on, risk off type thing. You could say, look, I've got lots of confidence about this forecast, you should take a decision based on this. And there are other times where you might say, look, the models are saying a whole variety of things. You could pull the data out of one of those models and AI could tell you go ahead and do this. But I've got the experience of seeing this a few times and when the models don't match up like this, you could be doing something which is a bit silly because there's a bit more context than just say what one model is saying. Is that, is that a sort of fair way of describing how people might use it?
- [25:03]
B
Yeah, I mean you learn what models tend to do better with certain things. For an example, just recently, I mean we've had some quite hot conditions recently and models tend to do a little bit of a shoddy job around this time of year. They sort of under predict they haven't quite got to that level in terms of the maximum temperatures that are coming through. But some of the AI models are just a little bit quicker at learning around that time. And so we saw quite consistently that the AI models are about a degree warmer around the hot spell that we've literally just seen and so we were saying, okay, actually, the traditional EC model might not ever actually get there, but it's likely that outturn, which is what we call delivery time, that actually you'll see something much closer to the AI. So you sort of get the experience of which one does tend to do better in which scenarios, and so you start leaning towards that.
- [25:49]
A
The EC model, that was your. That was the kind of one of the ways that you were doing it.
- [25:54]
B
Yeah, that was just an example. I mean. Okay, okay, yeah. So you have a traditional ec, you have an AI ec. So quite often comparing the two against each other. EC stands for ecmwf. Sorry.
- [26:04]
A
Oh. Which is the European center for Medium Weather Forecasts.
- [26:09]
B
Thank you.
- [26:09]
A
Thank you. Okay. All right, look, I'm not up to speed on my weather acronym, so you have to help me out. And it feels like an obvious place to get this conversation to is around climate change. It feels like the interest in weather is kind of forever growing because people want to be able to track the impacts of climate change. Is your job getting harder as years go by? And in addition to that question, are you also seeing a lot of the context that you've learned over, say, the last 10, 20, 30 years, you're kind of having to put in the bin because the new weather system is different to the old one?
- [26:45]
C
I'm happy to take this question, being the slightly older member of the team. So I've been forecasting, forecasting for just over 20 years. And one way in which we do forecasting, especially long range, is we look at historical data. So a little bit like AI, as in we look back at what weather patterns have done before, and there's definitely some evidence that the climate is changing rapidly. I mean, for example, we've seen marine heat waves where the seas become 3, 4, 5, 6 degrees warmer than normal, normal than the baseline, just off the coast of Europe. And even in the next week, the Mediterranean is going to become about 28 or 29 degrees. So there's these rapid warmups of the sea. And we feel that looking at historical data is still valuable, but perhaps there are these cases now where we can only look at a sort of more recent subset of years. And of course, if you reduce your sample size of years that you're comparing, that reduces your usefulness of those skills. So that's called analog forecasting using historical data. Climate change in Western Europe's warming up faster, I think, than almost anywhere on the planet over the last couple of decades. So when it's hot in Western Europe, we tend to see exceptional differences to the climate. For example, next week in France it could be into the 40s. We're getting these kind of 40 degree events in France more and more often there. So it's affecting our forecasting, it's affecting our ability to maybe use historical data. But we're getting massive interest from clients on these events and particularly for power.
- [28:06]
A
Right. So for lots of people listening, power generation generally needs water to cool it down. It's kind of how a lot of reactors work. And if you see 40 degree temperatures in France, then you can get to the point where rivers get quite hot. You're only allowed to let rivers get so hot before you kind of cook everything in them.
- [28:22]
C
Yeah.
- [28:22]
A
And so if you, if you have this like strong. We're kind of too early in the year for this because of a lot of the rivers will be coming from the Alps and so that water will still be quite cold. But as you get later into the year, if that water gets hotter and hotter and hotter, you end up not being able to run those power plants as much as you would like to. I remember a few years ago we had something in Germany where the river level got so low that it was difficult to run barges up. And so like there's all these kind of like second order impacts as well that starts coming.
- [28:50]
C
I think I'll probably hand to Emma here because she's done a lot of work on this or even next week we're getting quite excited about what may happen.
- [28:55]
B
So hydro is one of my area of interest.
- [28:59]
A
Okay.
- [29:00]
B
So with the issue around river levels and things, I mean we've built, with help from these guys, have built a sort of river temperature model for France. And it is a very. We were saying earlier about one of the misconceptions of the weather is that it is a multivariate problem. It's not just, I don't know, the temperatures are going to 40 degrees and so your temperatures and your rivers are going to go up. I mean they will. But obviously if your river levels are quite high, it's going to be slower to react to whether they're really low and then that warming will be more rapid. And so you can get that disparity coming through. But I mean at the moment we are in a situation where overall over the Alps you did see below normal snow over the season and so that melt water coming through is actually slightly lower than usual. We've seen some dry up spells coming through in towards France. And so actually if you look at the levels now, we're not actually that dissimilar to where we were in 2022, which is that year that you referenced, which was a bit shoddy on those levels. And so, yes, this heat spell coming through, those levels are low. And so you could still see that reaction and as early as next week. I mean, edf, who are the transmission system there, are starting to warn of risk, particularly around and towards that sort of southwestern side and on the Lyon river around there, that kind of zone, because you are seeing those temperatures coming through those rivers are lower. And it's a similar story in the Rhine at the moment.
- [30:19]
A
Sorry. So edf, a warning of nuke curtailment.
- [30:23]
B
So that's where you see, as you're saying you have to turn the reactors off because you just don't have that cooling water availability.
- [30:28]
A
Okay.
- [30:29]
B
So you need a certain amount.
- [30:30]
C
About this time last year we had a similar event actually late June 25th. So when the river level drops, it doesn't take as much energy to heat up that river. Of course, there's less. Less volume. And then, yeah, once it gets to the threshold. So we're thinking, yeah, next week we'd agree with the edf.
- [30:44]
A
Yeah.
- [30:44]
C
Now they were predicting a little bit of an earlier onset of threat for curtailment, but now they're in line with our initial view.
- [30:50]
A
So very good.
- [30:50]
C
But we're happy with our initial forecast.
- [30:52]
A
I'm sure the weather team and EDF are listening. And Emma, sorry I interrupted you. You were going to the Rhine, I think.
- [30:59]
B
Yeah. I mean, it's a similar story in the Rhine. You had that less snowmelt than usual. We've had less heavy precip coming in towards those zones into the Rhine. And so actually coming around onto where, for example, the Korb or Duke Rut Roh. I'm probably saying this completely wrong. I apologize to any Germans around, but those levels are falling. I mean, at the moment we're talking about coming down to a meter threshold which will start be pushing you, I mean, towards those excess. Right. Charges. And so again, 2022, we're not too far away from that. We could see a similar dive where we do come below those sorts of fresh values for more excessive.
- [31:33]
A
And those threshold values, they stop power stations pulling in the water. Or is that more about freights?
- [31:38]
B
For the Rhine, it's more freight charges. So you've got four sites coming through and they control the sort of barge that was going through. So there's different levels and there's four zones. Each one has a slightly different system in terms of level amounts that they allow free. But if your barge is basically too heavy and the river level's not enough, you're going to sink it. So it's sort of. If you hit a certain level, they'll start reducing the amount you can put on your barge. They'll start decreasing the traffic to stop that clogging up.
- [32:08]
A
We've all learned about how bad it can be when ships like the Evergreen crash into the Suez Canal halfway through. So lesson learned there. I've got one final question for you, Emma. I think I'm coming to you with this, but before I do that, do you ever find that like on a personal note, do you ever find that you're like on holiday and you're so walking past the Rhine and you're like, I could tell you that The Rhine is 8 degrees right now and it's like a meter. Like, do you ever find like, is this something that you find in your day to day life?
- [32:36]
C
Personally, the sea temperature is massive for me. Okay.
- [32:38]
A
You get a text, someone's like, I'm off to Crete next week, could you let me know, you know if it's
- [32:42]
C
going to be A, because how are you thinking about your swimming and B, you're thinking about is this going to produce loads of rain for the Alps in the autumn once the sea's to 28 degrees. So yeah, sort of two things were
- [32:51]
A
going on in that I would love to think you sort of like have this moonlight role for all of your friends where you're kind of also doing like a personal weather forecast. You know, you're getting, getting married next, next, next spring. Could you give me the long range? The long range forecast?
- [33:03]
C
There's nothing more stressful than a family forecast.
- [33:06]
A
Oh, okay, okay.
- [33:07]
C
Forget the client forecast.
- [33:08]
A
Okay. Okay, good to know. All right, Emma, then I'm going to come to you. What is a contrarian view you hold about weather modeling and energy systems?
- [33:15]
B
Yeah, so mine is to do with sort of headline weather. So for example, a really good example is something like the polar vortex. So that's something about higher levels which propagates down. And quite often there is this thing that we call a dinosaur chart. But it's sort of like a curve. So it basically averages over northern hemisphere and you take these zonal winds and you average over from the sort of northern hemisphere up and it basically looks like that. It's like the back of a stegosaurus.
- [33:45]
A
The shape of it looks a bit like a dinosaur. Okay, what's on that chart? What are the axes?
- [33:50]
B
Sorry, so it's zonal Wind. So that's your. And then you've got time down here. And so when you see that dip below zero, that means that the zonal winds have reversed, which is quite often. That's. You might have. Well, you might not have. But in the headlines that's called a sudden stratospheric warming and ssw, which to most traders then raises quite strong alarm bells of crazy cold in the winter. And so they panic and think, oh my gosh, we got this coming. But the issue is that, as I said, it's averaged over the entire northern hemisphere. It's not telling you the whole story. And it brings back to that common misconception question earlier, that things are multivariate. So yes, it could be dropping below zero. That could actually bring your cold frat, if it's sort of located and coupled in that zone, more so towards the us and so that then fires up our jet stream, which then brings more sort of low pressure to the north and hotter conditions over Europe. And so actually you've completely gone with the wrong side because you've seen this SSW and panicked, basically. Or you could see it just doesn't couple at all. So was it a couple of years ago it was record high in terms of the zonal winds, which normally means breezy and nice sort of windy conditions for northern parts of Europe. It just wasn't coupled. So you didn't get that effect. And actually we had. That was our Dunklefelter year.
- [35:09]
A
Okay.
- [35:09]
B
So the complete opposite.
- [35:11]
A
So your contrarian view is that the, that the headline, the headline piece, that sometimes a trader or like people who care about the weather, like from a, like as their job, they get really attached to this kind of concept of the main thing. But actually the concept of the main thing is potentially really overstating it. And that actually what might happen is almost the exact opposite because it may go slightly in a different direction, which means that kind of, to the earlier point, it nudges a system, let's say, get good jet stream coming back into Europe. So instead of being sort of record coals, record cold period, you're actually going to get a sort of much milder period in winter.
- [35:46]
B
Yeah. So it's like as soon as we hit September, I would willingly put money on the fad. The clients will start asking us SSW risk and things like this. I would happily put money on that fact. And you could say, I mean, last year we were saying, yes, it's above normal climatologically. However, that doesn't tell you the whole story. And you have to really hit that home because they're all going away thinking, oh, it's going to be a nice cold winter, we're going to get on this now. But actually that might not be the case. And it's not until sort of close to the time that you start seeing things like could we see that coupling coming through and so it being impactful or could that be more.
- [36:17]
C
I think it all links back to something called the Beast from the east, which you may have heard of in March 2018, where there was one of these, these sudden stratospheric warmings, a very powerful one that had almost a textbook response. So we had this event that Emma was talking about in the atmosphere higher up. The stratosphere is the next level up from the troposphere where we all the weather occurs, but it can connect down and couple down. And that did couple down and we ended up with a huge easterly wind and the gas price shot up, massive demand for heating. And of course everyone thinks about that event of could it happen again? And of course since then we've never seen anything quite as dramatic, but there's always that threat. So I think that's what makes traders trigger when they hear this term ssw,
- [36:58]
A
people have that sort of like more near term bias around it. They remember the Beast from the east, it also rolls off the tongue very nicely.
- [37:05]
C
But of course our job is to present data and sort of say, well, the Beast from the east was actually the coldest impact we've ever had from an SSW. In the last 40 years. We've never had anything quite as dramatic as that. So we have to show that these opposite cases, the US may end up with some of the cold, we may go really mild and it's just sort of being aware of the dynamics of the atmosphere. So we do a lot of research on that and to help our clients,
- [37:27]
A
I hope it's come across. I've loved this episode. I've learned a huge amount and I feel, as I said right at the start, we could have gone on for way longer, but we have to draw it to a close. Emma, Matt, thank you very much for coming on. You've been fantastic guests. I've learned a huge amount about weather. I'm sure our listeners have too. And we're looking out for the record El Nino later on this year.
- [37:45]
C
Great, thanks Ed.
- [37:45]
B
Thank you for having us.