WEBVTT 1 00:00:04.759 --> 00:00:08.320 With Laurent's segele end from London and Gerard Reed from Berlin. 2 00:00:08.839 --> 00:00:10.599 This is redefining Energy. 3 00:00:11.320 --> 00:00:11.599 Today. 4 00:00:11.640 --> 00:00:14.439 On Redefining Energy, we're going to talk about the weather. 5 00:00:14.279 --> 00:00:20.199 Lauren, yes, job, it's September now, Simmer's gone for Lisa, 6 00:00:20.640 --> 00:00:24.519 Lisa in particular. As for it, I will sing a 7 00:00:24.600 --> 00:00:27.519 song that that's going to be at the conclusion. But 8 00:00:27.640 --> 00:00:29.280 first of all, from my partner. 9 00:00:29.480 --> 00:00:32.479 A b Looco Energy is Europe's premier Lisa of ten 10 00:00:32.560 --> 00:00:36.880 foot container mobile batteries built in Europe with COTL best 11 00:00:37.000 --> 00:00:42.799 LFP cells. A Bloco Energy serves fourteen European countries, including France, 12 00:00:43.320 --> 00:00:46.960 Germany and the UK. A Bloco's batteries can be leased 13 00:00:47.000 --> 00:00:50.119 for any duration between six weeks and six years, and 14 00:00:50.159 --> 00:00:53.520 they are monitored by the Dutch award winning platform school 15 00:00:54.240 --> 00:00:57.719 a Bloco Energy. Make your life easier, make your business 16 00:00:57.719 --> 00:01:02.359 more flexible. Back to the show, when you're going to 17 00:01:02.399 --> 00:01:04.840 sing the song like me, because nobody will listened to 18 00:01:04.920 --> 00:01:06.840 the podcast again if they hear my voice singing. 19 00:01:06.879 --> 00:01:12.040 So actually, I just want to say. What we're not 20 00:01:12.079 --> 00:01:14.719 going to do is talk about the weather across Europe, 21 00:01:14.840 --> 00:01:17.840 which has been pretty crazy over the last few months. 22 00:01:18.319 --> 00:01:22.280 But we're really going to talk about the importance I've 23 00:01:22.280 --> 00:01:25.959 been able to predict the weather for anybody that's running 24 00:01:26.079 --> 00:01:29.439 energy assets across the world. This is really becoming critical. 25 00:01:30.040 --> 00:01:34.799 So there is this new sector called the weather forecasting service. 26 00:01:35.400 --> 00:01:38.439 It's a market to sizes about three billion a year 27 00:01:39.040 --> 00:01:42.280 and it's going extremely fast. Now. Of course, you've got 28 00:01:42.920 --> 00:01:49.840 sat de lights, new tech droans, new application, better granularity, 29 00:01:50.159 --> 00:01:55.000 use of AI, lots and lots of incredible innovation happening. 30 00:01:54.920 --> 00:01:56.879 Which is why we sort of talked, listen, let's go 31 00:01:56.920 --> 00:02:00.000 and get an expert in. So we decided to invite 32 00:02:00.120 --> 00:02:04.680 Martin Fengler, who's the CEO and founder of meteo Matics, 33 00:02:04.760 --> 00:02:07.959 which is a Swiss based weather service company. 34 00:02:08.360 --> 00:02:12.360 Yeah. Meteumatics is part of those ten ish startup which 35 00:02:12.520 --> 00:02:17.199 emerged in the past decade and competing, of course against 36 00:02:17.439 --> 00:02:20.719 the legacy public agency, but with a very different approach, 37 00:02:21.479 --> 00:02:25.639 very sharp. So yeah, it was a very interesting interview. 38 00:02:26.520 --> 00:02:35.639 One thing I need to say, it's very geeky. Surest Apron, Okay, 39 00:02:35.759 --> 00:02:39.120 let's listen to detail. Martin. Welcome to the show. 40 00:02:39.439 --> 00:02:40.639 Thank you soon, watching me. 41 00:02:41.039 --> 00:02:42.840 Great to have the weather man here. So I mean, 42 00:02:43.080 --> 00:02:45.080 what i'd like to do is maybe kick off by 43 00:02:45.159 --> 00:02:47.879 just asking just talk a little bit about the importance 44 00:02:47.919 --> 00:02:50.879 of the weather and the ability to predict the weather 45 00:02:51.000 --> 00:02:54.360 in the world of energy that were myself and RNA 46 00:02:54.520 --> 00:02:57.080 are living and working in every day well. 47 00:02:57.479 --> 00:03:01.280 Weather Yes, of course a huge impact on the energy industry. 48 00:03:01.599 --> 00:03:05.560 So you have, for instance, demand forecasts that are typically 49 00:03:05.759 --> 00:03:10.360 related to temperature, maybe radiation and wind speed, which is 50 00:03:10.439 --> 00:03:13.159 driving demand. But on the other hand, you have all 51 00:03:13.199 --> 00:03:17.159 these renewable energies like wind power, solar power, hydropower, and 52 00:03:17.560 --> 00:03:22.680 precise knowledge about the wind and the solar radiation, for instance, 53 00:03:22.719 --> 00:03:27.000 helps to predict the productions of those assets and ultimately 54 00:03:27.000 --> 00:03:31.120 helps you plan production much better with all the consequences 55 00:03:31.439 --> 00:03:33.479 or the different stakeholders in the energy sector. 56 00:03:33.680 --> 00:03:36.400 You've been the weather forecasting business for quite some time. 57 00:03:36.719 --> 00:03:40.080 Can you talk about the recent technological innovation that we've 58 00:03:40.080 --> 00:03:42.080 seen the PUS say fifteen twenty years. 59 00:03:42.240 --> 00:03:45.080 The weather industry has been actually quite old, so there 60 00:03:45.080 --> 00:03:48.879 have been players in the market for several decades. Especially 61 00:03:49.039 --> 00:03:51.680 national met services have been around for maybe on the 62 00:03:51.800 --> 00:03:52.879 years or even longer. 63 00:03:53.120 --> 00:03:53.879 But all of. 64 00:03:53.800 --> 00:03:57.719 Them have in common that they try to forecast weather. 65 00:03:58.199 --> 00:04:01.680 But it's a bottom line solving a really hard and 66 00:04:02.159 --> 00:04:06.599 challenging physical problem. So there are law of physics. Now, 67 00:04:06.599 --> 00:04:10.840 there are Stokes equations coup into feminodynamic equations, and it's 68 00:04:10.960 --> 00:04:14.879 really hard numerics that you've been in hands of National 69 00:04:14.879 --> 00:04:18.199 med Services. They had the budgets to purchase a supercompute 70 00:04:18.199 --> 00:04:21.480 power needed to run these models. But as we all know, 71 00:04:21.920 --> 00:04:24.519 the weather forecasts as we have them as if today 72 00:04:24.680 --> 00:04:28.839 have their flaws. So if you think of forecasts low 73 00:04:28.879 --> 00:04:33.439 stratus or even storms, it's quite evident that there's something 74 00:04:33.519 --> 00:04:37.480 missing when it comes to quality, and when you have 75 00:04:37.519 --> 00:04:40.759 a close look to it, then you'll start to realize, Hey, 76 00:04:41.160 --> 00:04:45.279 models that are used typically and computations are just to 77 00:04:45.480 --> 00:04:49.759 cause to resolve these small scale phenomena, and often enough 78 00:04:49.800 --> 00:04:52.160 you don't have a clue how the weather is right now, 79 00:04:52.319 --> 00:04:56.600 which is for layman typically irritating because you can just 80 00:04:56.639 --> 00:04:58.399 look out of the window and you see all the 81 00:04:58.399 --> 00:05:00.759 weather is outside. But of course the doesn't mean that 82 00:05:00.839 --> 00:05:04.279 the computer knows about it and knows exactly about the 83 00:05:04.319 --> 00:05:07.480 wind speed, wind direction and temperatures of force to do 84 00:05:07.519 --> 00:05:10.920 the weather forecast. And this is how matematics was born. 85 00:05:11.680 --> 00:05:14.800 We believe that it's time to do better and to 86 00:05:15.480 --> 00:05:19.240 do more high resolution weather forecasts and also to improve 87 00:05:19.279 --> 00:05:22.399 the initial state. And yeah, we have been working on 88 00:05:22.439 --> 00:05:25.360 this now for the last thirteen years by different means. 89 00:05:25.879 --> 00:05:28.439 So Martin, maybe just to dig into exactly what you're 90 00:05:28.480 --> 00:05:31.560 doing differently than national weather agencies. 91 00:05:32.480 --> 00:05:36.279 So national weather agencies are running their own models, but 92 00:05:36.480 --> 00:05:40.839 because it's their job from xpair is mainly to do 93 00:05:40.879 --> 00:05:44.439 it inside a specific country or to the country border. 94 00:05:45.160 --> 00:05:48.040 You rarely see them running really global models. There are 95 00:05:48.240 --> 00:05:51.439 only a few global models available and what we are 96 00:05:51.480 --> 00:05:54.720 doing is that we try to disrupt this. Next time 97 00:05:54.720 --> 00:05:58.519 we fly from Zurich to London, you need to recall 98 00:05:58.600 --> 00:06:01.839 that you're crossing maybe three different high resolution models from 99 00:06:02.000 --> 00:06:05.240 three different national met services. And this is something that 100 00:06:05.279 --> 00:06:08.920 we do differently. So we run, for instance, a Pan 101 00:06:08.920 --> 00:06:13.720 European and US high resolution model, so that is all 102 00:06:13.839 --> 00:06:16.839 unique and the resolution is way higher than what you 103 00:06:16.879 --> 00:06:20.120 can get as today from the national met services. So 104 00:06:20.160 --> 00:06:25.000 that's one piece a larger domain, higher resolution, and you 105 00:06:25.120 --> 00:06:28.759 need to invest also a lot more in getting data 106 00:06:29.000 --> 00:06:34.600 into your weather model that are currently not assimilated. Stimulated 107 00:06:34.759 --> 00:06:38.879 means that those weather information that you get from certain satellites, 108 00:06:38.959 --> 00:06:42.720 that you get from other measurement sources that might have 109 00:06:42.800 --> 00:06:45.720 also some proprietary nature, such as the meteor drones that 110 00:06:45.759 --> 00:06:48.439 we are working on whether whether the National met Service 111 00:06:48.480 --> 00:06:51.839 don't have access to and so you can improve the 112 00:06:51.879 --> 00:06:55.399 initial state, gets you closer to the current weather conditions 113 00:06:55.399 --> 00:06:58.480 and helps you also to forecast much better than the 114 00:06:58.600 --> 00:06:59.800 National met Services can do. 115 00:07:00.439 --> 00:07:03.600 Weather forecasting is the question of when and where. So 116 00:07:03.920 --> 00:07:07.439 when means how many days is the current state of 117 00:07:07.519 --> 00:07:10.720 the art, and compare with you a few years before? 118 00:07:11.319 --> 00:07:13.720 And where is how granular you can be from a 119 00:07:13.800 --> 00:07:17.040 geographical point of view? How do you compare? 120 00:07:17.560 --> 00:07:21.560 In our services, we provide our customers for any arbitrary 121 00:07:21.639 --> 00:07:24.319 that long weather data from the service up to an 122 00:07:24.399 --> 00:07:28.439 altitude of maybe twenty five kilometers. And this is something 123 00:07:28.600 --> 00:07:33.720 often enough that relies on publicly available or commercially available 124 00:07:33.920 --> 00:07:37.240 data such as from the European Center of Medium Range 125 00:07:37.279 --> 00:07:40.920 Leather forecasts from the UK Met Office from and this 126 00:07:41.040 --> 00:07:46.040 helps us to harmonize the existing information and to provide 127 00:07:46.040 --> 00:07:50.959 for instance also seasonal forecasts, maybe even climate projections or 128 00:07:51.319 --> 00:07:55.519 certain historical data. What Mediumatics is focusing on on top 129 00:07:56.000 --> 00:07:59.759 is that we run these high resolution models across Europe 130 00:07:59.759 --> 00:08:03.120 and yes, and for some if you other regions as well, 131 00:08:03.560 --> 00:08:06.639 but they are mainly focusing on the intra day day 132 00:08:06.680 --> 00:08:10.360 ahead and maybe three days ahead. So it's really more 133 00:08:10.439 --> 00:08:14.160 sort of a short term Betther model which helps us 134 00:08:14.240 --> 00:08:18.279 to address many applications that you come across in industry. 135 00:08:18.879 --> 00:08:21.399 Can I ask you to dig into this idea resolution 136 00:08:21.439 --> 00:08:24.160 Because I'm managing a wind farm, all I want to 137 00:08:24.199 --> 00:08:26.920 know is how strong the wind is going to be all. 138 00:08:26.800 --> 00:08:27.839 Over a period of time. 139 00:08:28.600 --> 00:08:30.399 That is that not all that I'm interested in. 140 00:08:31.480 --> 00:08:36.000 Someone who runs a wind turbine different aspects to it. 141 00:08:36.360 --> 00:08:39.440 So you have a rot or diameter that spands out, 142 00:08:40.000 --> 00:08:43.399 say fifty meters above the service to say one hundred 143 00:08:43.440 --> 00:08:47.039 and fifty meters, so it requires already wind information in 144 00:08:47.120 --> 00:08:51.759 different levels, but also temperature and air pressure that changes 145 00:08:51.960 --> 00:08:56.120 over time, and all of this is affecting the production 146 00:08:56.279 --> 00:08:59.240 of the wind speed. And it's quite obvious also the 147 00:08:59.240 --> 00:09:02.639 wind direction la role because of maybe shadowing or channelization 148 00:09:02.759 --> 00:09:06.279 effects or sea breeze. With this example, you can already 149 00:09:06.759 --> 00:09:11.200 see how many aspects come into the game when it's 150 00:09:11.240 --> 00:09:15.120 about precisely forecasting what's happening at turbine level. And if 151 00:09:15.159 --> 00:09:19.000 you think of multiple turbines in the park then maybe 152 00:09:19.279 --> 00:09:23.679 neighboring wind turbines or affecting each other because of wake effects. 153 00:09:24.360 --> 00:09:27.519 Okay, so this is this whole area of wind theft 154 00:09:27.600 --> 00:09:30.440 which we're beginning to hear in the offshore wind area. Right, 155 00:09:32.159 --> 00:09:35.080 Explain that to us and give us your view on this, 156 00:09:35.200 --> 00:09:37.399 because it seems to be a little bit strange. I go, well, 157 00:09:37.879 --> 00:09:40.600 I mean, it's your own fourth for wind turbines beside 158 00:09:40.600 --> 00:09:43.039 another wind turbine. Right, It's obvious if the wind comes 159 00:09:43.120 --> 00:09:44.600 in the west, you're going to steal it, or not 160 00:09:44.639 --> 00:09:46.799 steal it, you're going to change the weather pattern. 161 00:09:46.879 --> 00:09:47.080 Right. 162 00:09:47.799 --> 00:09:50.799 That What surprises me most is that people are surprised 163 00:09:50.840 --> 00:09:53.960 by it because the turbines are actually just doing what 164 00:09:54.000 --> 00:09:57.960 they've been designed to do, namely to collect energy out 165 00:09:58.000 --> 00:10:00.679 of the wind. So when you take the wind energy 166 00:10:00.759 --> 00:10:04.120 out of it, then now you have pretty calm winds 167 00:10:04.279 --> 00:10:06.840 behind on the lee of the such a wind farm. 168 00:10:07.799 --> 00:10:10.919 And we call these vake effects, and they can actually 169 00:10:11.399 --> 00:10:15.919 extend over large distances. So again layman might assume it's 170 00:10:16.200 --> 00:10:19.919 maybe just two hundred meters or one mile something like 171 00:10:19.960 --> 00:10:24.279 this that has affected it. But depending on the weather conditions, 172 00:10:24.559 --> 00:10:30.120 those wake effects can effect downstream wind and weather patterns 173 00:10:30.159 --> 00:10:34.279 in twenty thirty forty kilometers distance. And if you think 174 00:10:34.279 --> 00:10:36.559 of some of the wind farms that have been already 175 00:10:36.600 --> 00:10:40.159 built in the North Sea, for instance, and look into 176 00:10:40.200 --> 00:10:43.159 the plants that say how they want your expanses to 177 00:10:43.320 --> 00:10:46.840 till say twenty to fifty, you will see lots of 178 00:10:47.000 --> 00:10:51.440 discussions and in your future around this topic. So the 179 00:10:51.639 --> 00:10:54.120 UK wind farms in the North Sea might be here 180 00:10:54.600 --> 00:10:59.200 in favor because they're in Europe. We have often these 181 00:10:59.200 --> 00:11:05.200 westerly conditions and they getting the wind first. The others 182 00:11:05.240 --> 00:11:06.960 need to see what's left over. 183 00:11:07.600 --> 00:11:12.200 So you're using somebody else satellite. So your expertise is 184 00:11:12.200 --> 00:11:16.519 to possess billions of data all the time and compare 185 00:11:16.600 --> 00:11:19.600 with the past and everything. And also you're using, and 186 00:11:19.639 --> 00:11:22.679 that's proprietory what you call your drones. So can you 187 00:11:22.720 --> 00:11:25.399 explain a bit what they are doing and what additional 188 00:11:25.519 --> 00:11:28.120 quality they bring to your results. 189 00:11:28.679 --> 00:11:32.159 So satellites have been around and weather for maybe fifty 190 00:11:32.279 --> 00:11:36.039 years already, and here we are mainly typically in whether 191 00:11:36.080 --> 00:11:40.919 we discuss geostationary satellites. Satellites operating at an altitude of 192 00:11:41.559 --> 00:11:46.080 maybe thirty eight thousand kilometer distance to the airth surface 193 00:11:46.240 --> 00:11:51.080 and really well designed and well engineered instruments. And then 194 00:11:51.120 --> 00:11:54.639 you have also low earth orbiters typically operating in four 195 00:11:54.679 --> 00:11:59.720 hundred kilometers distance. And again these are typically operated and 196 00:12:00.000 --> 00:12:04.519 owned by government entities such as humids such in Europe 197 00:12:04.519 --> 00:12:07.720 and in NOA in the US, and all of them 198 00:12:07.759 --> 00:12:13.440 haven't common that they typically use optical sensors in different 199 00:12:13.480 --> 00:12:17.720 spectral channels and this gives you some great pixel value. 200 00:12:18.679 --> 00:12:22.480 And this great pixel value that you detect can be 201 00:12:22.600 --> 00:12:27.919 related to some radiation input and after some calibration procedure, 202 00:12:27.960 --> 00:12:32.840 for instance, you can determine how a certain amount of water, 203 00:12:32.960 --> 00:12:38.360 vapor or a cloud is then monitored with such an instrument. Now, 204 00:12:38.399 --> 00:12:41.720 the problem is these instruments is that they don't give 205 00:12:41.759 --> 00:12:45.840 you any direct measurements of the air temperature, the windspeed, 206 00:12:45.919 --> 00:12:52.039 in direction, air pressure, nothing even moisture is not directly measured. Therefore, 207 00:12:52.279 --> 00:12:56.960 those instruments are typically not that accurate. They give you 208 00:12:57.159 --> 00:13:01.399 a very strong signal and a super important for in 209 00:13:01.440 --> 00:13:04.200 the microather focus, so I'm not taking this into doubt. 210 00:13:04.960 --> 00:13:08.200 But when it comes to very local phenomena, especially those 211 00:13:08.240 --> 00:13:11.440 that are triggered by air masses close to the air surface. 212 00:13:11.519 --> 00:13:14.720 So in the first say three or four kilometers, then 213 00:13:14.799 --> 00:13:18.759 these instruments are typically not that accurate, and this is 214 00:13:18.799 --> 00:13:22.960 why whether stations are still important. But of course also 215 00:13:23.080 --> 00:13:26.759 radio signs and some other remote sensing instruments or even 216 00:13:26.919 --> 00:13:30.679 aircraft data, they are all super relevant because they give 217 00:13:30.720 --> 00:13:34.320 you direct measurement readings. And with our drones, we are 218 00:13:34.399 --> 00:13:38.519 actually replicating this and having a drone equipped with which 219 00:13:38.600 --> 00:13:42.320 logical sensors which gives us direct measurement readings. 220 00:13:42.960 --> 00:13:45.320 Martin, can I ask you about the economic side of it, 221 00:13:45.320 --> 00:13:47.039 because what I'm trying to get my head of if 222 00:13:47.080 --> 00:13:50.440 I was an energy trader or as at the end 223 00:13:50.480 --> 00:13:53.279 of the day, what I'm trying to do is make 224 00:13:53.320 --> 00:13:55.879 as much money as possible for as little risk as possible. 225 00:13:56.039 --> 00:14:00.320 So how does it turned into economics? That's really weather prediction. 226 00:14:01.559 --> 00:14:03.919 What we of course do is that we sell to 227 00:14:04.120 --> 00:14:09.240 energy and trading companies high resolution, high quality wind information 228 00:14:09.559 --> 00:14:13.440 and solar power information. Gives our customers some advantage that 229 00:14:13.480 --> 00:14:17.200 they can immediately leverage at the exchange. Now, coming back 230 00:14:17.240 --> 00:14:20.200 to the drone piece, for instance, it's not necessary that 231 00:14:20.240 --> 00:14:24.840 you have drones being installed on say every kilometer. It's 232 00:14:25.120 --> 00:14:28.679 absolutely not the case and the reason is, and this 233 00:14:28.799 --> 00:14:32.759 is where the secret source lies, is that you have 234 00:14:32.840 --> 00:14:37.759 a four dimensional data simulation. This means that every single 235 00:14:37.879 --> 00:14:42.320 profile that you do now helps you also with the 236 00:14:42.360 --> 00:14:47.080 next model initialization that you do, even though the measurement 237 00:14:47.159 --> 00:14:51.120 is one hour old, but the information is advacted with 238 00:14:51.200 --> 00:14:54.840 the wind. It's transported with a wind. Simple example, if 239 00:14:54.840 --> 00:14:58.159 you do, for instance, and a profile in Zurich, and 240 00:14:58.240 --> 00:15:01.600 it takes about an hour on aver that this air 241 00:15:01.679 --> 00:15:06.440 mass is reaching sangaland that is about seventy kilometers to 242 00:15:06.559 --> 00:15:09.879 the east of the ric. What it ultimately means, even 243 00:15:09.919 --> 00:15:14.080 if you do these soundings and a small amount of occasions, 244 00:15:14.279 --> 00:15:19.879 it helps you to make those sensings in many locations 245 00:15:20.639 --> 00:15:23.840 just because of the fact that the wind is transporting 246 00:15:23.919 --> 00:15:24.639 this information. 247 00:15:25.519 --> 00:15:28.200 Maybe just to change topic as slide, but when I 248 00:15:28.240 --> 00:15:31.679 listen to you, I can hear very clearly the benefits 249 00:15:31.679 --> 00:15:35.240 in the energy space. But I also think that if 250 00:15:35.320 --> 00:15:39.799 benefits in other areas. I can imagine rescue missions, military missions. 251 00:15:40.000 --> 00:15:42.759 You can go through a whole pile of areas where 252 00:15:43.000 --> 00:15:44.960 accurate weather prediction is important. 253 00:15:45.799 --> 00:15:49.960 Yeah, but that touches everything. I've been in this business 254 00:15:50.000 --> 00:15:54.039 now for nearly twenty years and I still learn at 255 00:15:54.120 --> 00:15:57.000 least once a week about the new use case. You 256 00:15:57.080 --> 00:16:01.240 have customers in the automotive industry such a Porsche may 257 00:16:01.279 --> 00:16:06.320 see this in w whatsoever. They all need weather for 258 00:16:06.519 --> 00:16:11.960 certain applications, save for entertainment in the car, for predictive maintenance, 259 00:16:12.000 --> 00:16:15.840 for autonomous driving. Just with these examples you already see 260 00:16:16.039 --> 00:16:19.960 how diverse weather applications can be. And this is just automotive. 261 00:16:20.639 --> 00:16:23.639 Then you have maybe aviation, where we work with companies 262 00:16:23.720 --> 00:16:27.440 like air Bus, with Lockheed, with Tahlis and others. And 263 00:16:27.519 --> 00:16:32.720 again there might be about aircraft engine optimization. Again you 264 00:16:32.799 --> 00:16:36.039 need high altitude weather to do that. And you can 265 00:16:36.120 --> 00:16:40.159 continue with all of those industries and it's super fascinating 266 00:16:40.919 --> 00:16:44.559 how diverse weather applications is, and it goes far beyond 267 00:16:44.799 --> 00:16:48.320 what we typically have when we watch the BBC Weather, 268 00:16:49.000 --> 00:16:52.000 which shows just the UK map and then a few 269 00:16:52.080 --> 00:16:55.159 items dropped on it. It's a lot more than what 270 00:16:55.240 --> 00:16:56.559 we typically look at. 271 00:16:57.480 --> 00:16:59.799 If I look at your client least you have six 272 00:16:59.840 --> 00:17:04.799 on clients, and some extremely prestigious like NASA. Do you 273 00:17:04.880 --> 00:17:07.480 know the application they're doing or you're just giving them 274 00:17:07.519 --> 00:17:09.440 THATA and you don't really know what they do with 275 00:17:09.519 --> 00:17:12.440 those data. Is there a kind of collaboration, how does 276 00:17:12.480 --> 00:17:12.880 it work? 277 00:17:13.559 --> 00:17:18.240 Often do you know about the application? In case of NASA, 278 00:17:18.279 --> 00:17:22.880 I can tell it's about supporting slide tests for some 279 00:17:23.200 --> 00:17:27.039 prototypes that they're looking at. And this is of course 280 00:17:27.319 --> 00:17:32.200 whether again it's affecting for instance, not just because of 281 00:17:32.279 --> 00:17:37.079 turbulence or maybe icing aircraft operations. But here, for instance, 282 00:17:37.160 --> 00:17:42.880 it's about sound propagation. So you think of ultrasonic plane 283 00:17:42.880 --> 00:17:47.160 and aircraft, then you have this sonic boom and the 284 00:17:47.279 --> 00:17:51.319 boom propagation is affected by weather again and it's again 285 00:17:51.359 --> 00:17:55.559 it's not just wind, but it's the combination of temperature 286 00:17:55.720 --> 00:18:02.200 and moisture which creates sound bending sometimes sometimes not, and 287 00:18:02.519 --> 00:18:04.440 just just depends on the better. 288 00:18:05.240 --> 00:18:09.720 How far away can you predict, because what's very very 289 00:18:09.759 --> 00:18:14.759 important in the energy business are mostly the dry and 290 00:18:15.000 --> 00:18:20.359 you know, not very windy winters versus the mild and 291 00:18:20.440 --> 00:18:23.200 windy winter, and that makes a lot of difference in 292 00:18:23.480 --> 00:18:26.799 you managed to get some long term you know, prediction 293 00:18:27.480 --> 00:18:30.359 when we arrive like in October, because this is really 294 00:18:30.400 --> 00:18:34.079 going to affect massively the all energy system for the 295 00:18:34.160 --> 00:18:35.000 coming months. 296 00:18:35.440 --> 00:18:39.960 So for these long term predictions we rely also on 297 00:18:40.039 --> 00:18:44.920 other sources and MWF. The European center of a medium 298 00:18:45.000 --> 00:18:48.920 range of the forecasting is certainly the gold standard if 299 00:18:48.960 --> 00:18:54.240 it comes to forecasts into the front week, three weeks, 300 00:18:54.279 --> 00:18:58.160 four weeks, and maybe seasonal forecasts. We utilize the output 301 00:18:58.599 --> 00:19:02.880 and refine it. We're adding, for instance, downscaling techniques, combining 302 00:19:02.920 --> 00:19:05.839 their cost grid data with high resolution terraer data and 303 00:19:06.160 --> 00:19:09.119 find unit. But at the moment we as me teenetics. 304 00:19:09.440 --> 00:19:13.279 We don't render the sort of long term forecasts. But 305 00:19:13.440 --> 00:19:16.920 of course our moonshot vision is to run a high 306 00:19:17.000 --> 00:19:21.759 resolution model, global one kilometer model one day, and we're 307 00:19:21.799 --> 00:19:24.400 making huge progress with this. We have been working on 308 00:19:24.440 --> 00:19:27.599 this for a couple of years now. When we start 309 00:19:27.640 --> 00:19:30.480 with it, people laughed at us. People told us, no 310 00:19:30.519 --> 00:19:32.960 one needs this, no one wants this. And now many 311 00:19:33.000 --> 00:19:36.880 of our competitors stopped laughing. I believe we have delivered 312 00:19:36.920 --> 00:19:39.880 now on a pan European one calumeter model on the 313 00:19:39.960 --> 00:19:41.960 US one calummeter model, and it's just a matter of 314 00:19:42.039 --> 00:19:44.319 time that we are going to deploy also a global 315 00:19:44.359 --> 00:19:45.279 one klumeter model. 316 00:19:46.000 --> 00:19:48.720 Martin, can I ask you something which is quite different, 317 00:19:48.799 --> 00:19:52.599 which is I'd really like to hear your opinion in 318 00:19:52.640 --> 00:19:56.160 and around what I call weather volatility, which is it 319 00:19:56.359 --> 00:20:00.640 seems that weather volatility is increasing. And listen, you've got 320 00:20:00.640 --> 00:20:03.200 all the data out there when you look at the weather, 321 00:20:03.519 --> 00:20:06.480 is our weather changing and if so, what does it 322 00:20:06.559 --> 00:20:07.400 mean for us? 323 00:20:07.839 --> 00:20:11.079 I think because this always depends on the region and 324 00:20:11.119 --> 00:20:14.319 all of this, but what we definitely see in the 325 00:20:14.400 --> 00:20:20.519 data is that the weather becomes more pronounced, more extreme 326 00:20:20.559 --> 00:20:25.799 from time to time. Climate change is introducing more radiation 327 00:20:26.480 --> 00:20:30.079 and this leads to more higher temperatures. Higher temperatures often 328 00:20:30.759 --> 00:20:35.480 can often lead to an increased moisture transport, and then 329 00:20:36.480 --> 00:20:41.240 you often see much higher intensity when it comes to storms, 330 00:20:41.880 --> 00:20:46.720 and there being some recent examples. I don't want to 331 00:20:46.759 --> 00:20:49.200 go that far and blame climate chain for that, but 332 00:20:49.519 --> 00:20:52.160 if you think of those extreme events that we just 333 00:20:52.599 --> 00:20:56.640 saw the other day in Texas or in Valencia, these 334 00:20:56.680 --> 00:21:01.720 are certainly extreme events that can be a consequence from 335 00:21:01.759 --> 00:21:02.359 this effect. 336 00:21:02.920 --> 00:21:06.920 Martin, you are part of the coupe of startups who 337 00:21:06.920 --> 00:21:10.759 are competing again national weather services. How do you see 338 00:21:10.799 --> 00:21:14.240 the balance between the two services going forward? 339 00:21:15.079 --> 00:21:19.440 We of course try to look into collaboration with National 340 00:21:19.440 --> 00:21:23.079 met Service wherever we can. But of course the natural 341 00:21:23.160 --> 00:21:25.720 question that let's call it the elephant in the room 342 00:21:26.119 --> 00:21:30.599 with National met Services whether the way how they operators 343 00:21:30.599 --> 00:21:35.440 of today is appropriate for the future. Meaning I envision 344 00:21:35.720 --> 00:21:42.119 that we see National met Services change over time and 345 00:21:42.599 --> 00:21:46.599 to become more like an authority like the Civil Aviation 346 00:21:46.759 --> 00:21:51.319 Authority for instance, or like a Ministry of Defense. And 347 00:21:51.400 --> 00:21:54.400 this means in terms of betther that they are a 348 00:21:54.519 --> 00:21:59.599 rulemaking organization and still have some operative element to it, 349 00:21:59.680 --> 00:22:04.000 like reassuring severe weather warnings, but all the tools and 350 00:22:04.160 --> 00:22:08.559 services they need to render the operations will be acquired 351 00:22:08.559 --> 00:22:11.640 from the market, as if today it is just some 352 00:22:11.680 --> 00:22:15.599 sort of historical accident that you still have the agencies 353 00:22:15.759 --> 00:22:20.000 that are doing all the programming themselves. But frankly, there 354 00:22:20.039 --> 00:22:23.000 is no reason to do so. If you can buy 355 00:22:23.039 --> 00:22:25.880 a fighter jet like an F thirty five through a 356 00:22:25.920 --> 00:22:30.559 bit process highly complex weapon systems in this case, when 357 00:22:30.559 --> 00:22:34.480 you can do this through a tenoring process, then frankly 358 00:22:34.519 --> 00:22:39.000 you can also purchase the weather model output from the market. 359 00:22:39.279 --> 00:22:43.960 You could also ask companies to run and operate weather 360 00:22:44.039 --> 00:22:47.759 station networks and just put some slas against it, and 361 00:22:47.839 --> 00:22:50.839 then you can actually buy all of this from the 362 00:22:50.960 --> 00:22:53.960 private sector from the market. And this is maybe something 363 00:22:54.000 --> 00:22:55.920 we are not going to see to change in the 364 00:22:56.559 --> 00:23:00.799 near future, especially not in Central Europe, but I expect 365 00:23:01.160 --> 00:23:03.880 that we are going to see this already in other 366 00:23:04.079 --> 00:23:07.759 areas of the world, that these services will be purchased 367 00:23:07.759 --> 00:23:08.359 from the market. 368 00:23:09.319 --> 00:23:12.039 Martin. When I look at the prediction, the market of 369 00:23:12.160 --> 00:23:15.079 weather forecasting is supposed to double the next ten years. 370 00:23:15.720 --> 00:23:18.400 So what technology do you have in mind and what 371 00:23:18.440 --> 00:23:21.039 type of result will we have in ten years that 372 00:23:21.119 --> 00:23:21.920 we don't have now? 373 00:23:22.559 --> 00:23:26.960 What I strong believe is that all the modern AI 374 00:23:27.160 --> 00:23:32.720 techniques that are discussed will find their way also into weather. 375 00:23:33.359 --> 00:23:36.319 I'm here a bit cautious because that at the moment 376 00:23:36.359 --> 00:23:38.680 also a lot of noise in the market around that. 377 00:23:39.440 --> 00:23:43.759 But there are certainly areas where AI methods can contribute. 378 00:23:44.079 --> 00:23:48.039 If you think of satellite data assimulation, for instance, which 379 00:23:48.079 --> 00:23:51.400 has always been a huge challenge in whether or if 380 00:23:51.440 --> 00:23:55.119 it comes to post processing. So for instance, if Jara 381 00:23:55.359 --> 00:23:59.119 is really interested in wind power output of a specific 382 00:23:59.160 --> 00:24:04.519 turbine civic location, then AI techniques helped to come up 383 00:24:04.559 --> 00:24:10.440 with even more precise focast because AI techniques can identify 384 00:24:11.119 --> 00:24:15.799 local phenomena that are not primetrized, not modeled, not reflected 385 00:24:15.839 --> 00:24:19.119 in another model. And this is where AI techniques can 386 00:24:19.160 --> 00:24:21.119 definitely add value. 387 00:24:21.319 --> 00:24:27.240 Well, Martin was geeky but interesting. The whole point of 388 00:24:27.279 --> 00:24:30.240 being experts like you is just to realize that it's 389 00:24:30.240 --> 00:24:32.279 not because you read three articles on the internet that 390 00:24:32.359 --> 00:24:36.359 you become an immediate international expert. So thank you very 391 00:24:36.440 --> 00:24:38.559 much for sharing your expertise with us. 392 00:24:38.839 --> 00:24:40.079 Thank you for having me here. 393 00:24:41.000 --> 00:24:43.079 Yeah, it's good, Thank you very much, Martin. 394 00:24:44.359 --> 00:24:49.039 Job. I found the conversation fascinating, and I'm glad we 395 00:24:49.119 --> 00:24:52.359 have this podcast so we can bring a real expert 396 00:24:53.279 --> 00:24:57.440 and not the Juninequgo expert like we are. Sometimes. 397 00:24:57.720 --> 00:24:59.240 I'm totally with ch and I know it's a little 398 00:24:59.279 --> 00:25:02.720 bit nerdy and technical, but it's really really critical. And actually, 399 00:25:02.720 --> 00:25:04.960 do you know the funny thing is I use her 400 00:25:04.960 --> 00:25:08.319 app all the time now by the prediction now right, 401 00:25:08.359 --> 00:25:09.119