WEBVTT 00:00.000 --> 00:04.000 In the Middle Ages, people were convinced that they were witches who 00:04.000 --> 00:06.680 let the destructive hail rain down on people. 00:07.820 --> 00:11.700 The consequences were not uncommon, a destruction of the harvest and 00:11.700 --> 00:12.500 the need for hunger. 00:13.180 --> 00:17.500 Today, the hail causes damage in the billions every year, increasing 00:17.500 --> 00:18.080 in tendency. 00:19.000 --> 00:23.400 But what actually happens in thunderstorms, when the rain suddenly 00:23.400 --> 00:27.620 turns into hail and tennis ball-sized ice crumbles on the ground? 00:28.390 --> 00:38.020 The aim of a research project at the Karlsruhe Institute of Technology 00:38.500 --> 00:43.520 When dark thunderstorm cells form over the Swabian Alps or the Neckar 00:43.520 --> 00:47.580 Valley this summer, the Karlsruhe thunderstorm hunters swarm out. 00:47.740 --> 00:51.260 With a lot of measuring equipment in their luggage, they chase the 00:51.260 --> 00:52.740 hailstorm -prone clouds. 00:52.740 --> 00:59.800 The aim is to observe the catastrophic phenomenon of hail up close and 00:59.800 --> 01:03.820 to improve the prediction so that, in the event of an accident, there 01:03.820 --> 01:08.080 is enough time to prevent the worst effects of the hailstorm. 01:09.200 --> 01:12.280 It's exciting and frustrating at the same time, because you always 01:12.280 --> 01:15.800 find yourself in the wrong place at the wrong time and may not catch 01:15.800 --> 01:16.300 the thunderstorm. 01:17.580 --> 01:22.080 We have a briefing in the morning, where we look at the current 01:22.080 --> 01:22.500 models. 01:22.820 --> 01:26.520 The current model of the German Weather Service is very helpful. 01:26.840 --> 01:29.020 It is set up anew every hour. 01:29.900 --> 01:32.700 The available data is assimilated into the model every hour. 01:33.960 --> 01:35.380 This is extremely helpful. 01:35.660 --> 01:38.580 First, we look at a convective situation, i.e. 01:38.580 --> 01:39.700 to expect a thunderstorm situation. 01:40.000 --> 01:43.180 Where is the highest probability of a thunderstorm? 01:43.180 --> 01:46.280 Then, of course, comes our long-term experience. 01:46.460 --> 01:50.580 For example, we know that Karlsruhe is not a good place to hunt for 01:50.580 --> 01:52.280 hail or severe thunderstorms. 01:52.500 --> 01:54.360 There are very few thunderstorms here. 01:54.500 --> 01:57.480 This is probably due to the current conditions, because we have a 01:57.480 --> 01:59.700 divergent current that triggers the thunderstorms. 02:00.120 --> 02:02.420 Then we set off and observe the radar. 02:02.720 --> 02:06.660 Then we refine our locations, where we go and where we go. 02:08.140 --> 02:10.660 Ideally, we want to position ourselves directly in the inflow area of 02:10.660 --> 02:13.820 the thunderstorm cell and start our measurements there. 02:13.960 --> 02:17.100 In his office on the campus north of the Karlsruhe Institute of 02:17.100 --> 02:21.200 Technology, the thunderstorm hunter Professor Michael Kunz can show 02:21.200 --> 02:24.020 true monster examples of hail grains. 02:25.340 --> 02:30.960 Eternally in plastic, they not only impress by their size of 10 cm and 02:30.960 --> 02:35.940 more, they are also provided with strange long bulges, which underline 02:35.940 --> 02:38.540 their extreme danger when they burst. 02:38.540 --> 02:45.360 If you cut open a larger hail grain, you can read its entire history 02:45.360 --> 02:48.000 of formation in the turbulence of the thunderstorm clouds. 02:49.060 --> 02:52.720 In the middle is the so-called embryo of the hail. 02:53.100 --> 02:57.460 This embryo is either a grain or it can also be a large frozen 02:57.460 --> 02:58.140 raindrop. 02:59.260 --> 03:02.760 This is the central growth unit and we have different layers. 03:03.100 --> 03:06.280 These layers are dry and wet growth. 03:06.280 --> 03:12.100 If undercooled drops hit such a hail grain and it freezes, heat is 03:12.100 --> 03:12.660 released. 03:13.040 --> 03:18.520 It is released and can ensure that the surface of the hail grain has 03:18.520 --> 03:19.760 zero degrees. 03:20.120 --> 03:23.580 This means that we have a water layer on the hail grain and the water 03:23.580 --> 03:26.340 layer ensures that water can penetrate into the pores created by the 03:26.340 --> 03:27.420 freezing process. 03:28.620 --> 03:31.200 Then we have a clear, transparent layer. 03:31.340 --> 03:34.400 If the temperature remains below zero degrees, it does not happen. 03:34.400 --> 03:37.660 Then it comes directly to the contact, to the spontaneous freezing. 03:37.860 --> 03:40.860 There are a lot of air inclusions in the structure, in the ice. 03:41.120 --> 03:43.120 This then appears as a milky layer. 03:44.000 --> 03:47.000 On the one hand, the shape of the hail grains is crucial. 03:47.280 --> 03:49.080 Here is a rather round hail grain. 03:49.480 --> 03:50.340 This one is more spiky. 03:50.620 --> 03:53.740 There we have a lot of protrusions and we can determine the growth. 03:54.360 --> 03:57.600 This layer structure also appears in the outer shape. 03:58.320 --> 04:01.880 We then scan the hail grains and cut them through and look at the 04:01.880 --> 04:02.440 layer structure. 04:02.440 --> 04:05.720 We just have the problem that we can't really look into the 04:05.720 --> 04:05.720 thunderstorms. 04:05.820 --> 04:09.620 With the radar, somehow, but then we have the problem with signal 04:09.620 --> 04:12.080 damping and we can't get all the information out of there either. 04:12.640 --> 04:16.720 But in such a hail grain, the entire life cycle of the thunderstorm 04:16.720 --> 04:17.020 cell is ultimately embedded. 04:17.180 --> 04:18.740 And it is necessary to decrypt it. 04:19.040 --> 04:24.060 What seems so simple, water droplets freeze and fall like ice from the 04:24.060 --> 04:28.560 sky, turns out to be a highly complex process. 04:28.700 --> 04:31.860 What we need to freeze is an aerosol. 04:31.860 --> 04:35.720 And these aerosols that are needed for the ice formation are very 04:35.720 --> 04:36.200 special. 04:36.500 --> 04:38.700 We call them ice nucleating particles. 04:38.920 --> 04:40.700 First of all, they need a certain size. 04:40.860 --> 04:41.680 They must not be soluble. 04:42.100 --> 04:44.500 Sea salt, for example, is not a good aerosol. 04:44.680 --> 04:46.060 Desert dust is not bad at all. 04:46.280 --> 04:47.620 Feldspat is actually the best. 04:48.240 --> 04:50.820 And the reason for this is that the grid structure must be similar to 04:50.820 --> 04:52.340 the grid structure of ice. 04:52.820 --> 04:55.940 If the grid structure is very different, then it is due to the 04:55.940 --> 05:01.280 bindings to offsets and ensures that when an ice molecule forms, it 05:01.280 --> 05:04.120 becomes unstable immediately and the bindings break apart again. 05:04.420 --> 05:06.940 That means a structure similar to that of ice is ideal. 05:07.460 --> 05:10.760 And there are very few of these aerosols in the atmosphere. 05:10.980 --> 05:11.800 How many are there actually? 05:12.580 --> 05:13.820 We also have measurements. 05:14.460 --> 05:18.400 There are about 10 or 100 particles per liter. 05:18.520 --> 05:19.320 That's all there is. 05:19.860 --> 05:23.320 As we said, we need these aerosols, these ice nucleating particles, 05:23.700 --> 05:24.680 for the ice formation. 05:24.820 --> 05:26.420 Now we have too few of these particles. 05:26.520 --> 05:27.140 What happens? 05:27.140 --> 05:31.780 In such a cloud we have few ice particles and a lot of water droplets 05:31.780 --> 05:34.260 with a temperature below zero degrees Celsius. 05:34.740 --> 05:36.560 We call them undercooled droplets. 05:37.140 --> 05:39.600 And they now accumulate on the few ice particles. 05:40.480 --> 05:44.820 The process is extremely fast, this accumulation, this adhesion or 05:44.820 --> 05:45.280 maturing. 05:45.760 --> 05:47.480 And then a grain particle is formed. 05:47.560 --> 05:52.580 And from the grain particle, then in an area below minus 5 degrees, 05:52.880 --> 05:56.040 where we have strong vertical velocities, strong turbulence. 05:56.040 --> 05:58.200 This is where the hail can form. 05:58.300 --> 06:04.100 So from a small grain particle, a hail cone in tennis ball size. 06:04.380 --> 06:06.400 Ultimately, it is a question of aging time. 06:06.800 --> 06:11.220 So how long is such a hail embryo, such a grain particle, and then the 06:11.220 --> 06:13.100 hail in the area where it can grow? 06:13.980 --> 06:16.340 The wider the wind, the better it is. 06:16.560 --> 06:19.040 And if it is very narrow, then the particles are easily transported 06:19.040 --> 06:20.080 out of the wind too quickly. 06:21.120 --> 06:21.980 Width is good. 06:22.580 --> 06:24.900 If the wind is too strong, then it is also bad. 06:25.460 --> 06:27.580 Because then they don't stay long enough. 06:27.660 --> 06:30.800 That means the particles have to stay in the edge area of ​​the wind. 06:31.300 --> 06:35.160 And then the hail can really get really big and strong. 06:35.760 --> 06:38.740 Beautiful is perhaps the wrong term in view of the damage. 06:38.840 --> 06:41.560 But hail can grow to a diameter of up to 20 cm. 06:42.960 --> 06:46.340 I learned at school that hail consists of several layers. 06:46.580 --> 06:49.020 And the layers are created by the fact that the hail grain is 06:49.020 --> 06:52.840 transported up, falls down, is transported up again, falls down. 06:53.560 --> 06:57.300 Everything we observe, what we measure, is not the case. 06:57.400 --> 06:58.880 This process cannot be entirely excluded. 06:58.980 --> 07:01.720 Then there may be times, but this is not the main growth process. 07:02.040 --> 07:05.240 So a hail grain is transported up once and then when the gravitation 07:05.240 --> 07:07.220 is too heavy, it just falls to the ground. 07:07.380 --> 07:11.940 As part of a research project with the fitting name LIFT, the 07:11.940 --> 07:15.340 Karsruher thunderstorm hunters bring a lot of measuring devices in the 07:15.340 --> 07:16.380 middle of the disaster. 07:17.660 --> 07:20.780 The researchers are concerned with the comparison between computer 07:20.780 --> 07:25.940 models, as well as the German Weather Service, and reality on site. 07:26.000 --> 07:28.620 The problem is that we can't really look into the thunderstorm cloud. 07:29.980 --> 07:31.280 We have such small probes. 07:31.620 --> 07:35.040 It has the shape of a hail grain, a diameter of 5 cm. 07:35.260 --> 07:37.880 And we try to get it into the upwind area. 07:39.080 --> 07:40.440 So with helium balloons. 07:40.580 --> 07:43.580 We then blow off the helium balloons when the probe is in the upwind 07:43.580 --> 07:44.080 area. 07:45.060 --> 07:48.440 Because then we get so-called Lagrange trajectories, or approaching 07:48.440 --> 07:49.940 Lagrange trajectories. 07:49.940 --> 07:54.860 This means that these tracks in the upwind area are ultimately the 07:54.860 --> 07:58.080 tracks that hail embryos and hail then complete. 07:58.660 --> 08:01.040 And we haven't been able to observe them like that before. 08:01.360 --> 08:02.500 This is actually the new thing. 08:03.100 --> 08:06.080 There is a GPS sensor, there is a temperature and humidity sensor. 08:06.260 --> 08:07.740 From this we then get the trajectories, i.e. 08:07.740 --> 08:08.140 the tracks. 08:08.740 --> 08:11.520 Temperature and humidity then provide us with the information about 08:11.520 --> 08:13.620 the conditions in the thunderstorm cloud. 08:13.720 --> 08:22.080 In addition, ground sensors are set up and drones are also used for 08:22.080 --> 08:22.080 this. 08:22.640 --> 08:25.980 Especially large hail grains are collected and brought to the 08:25.980 --> 08:26.640 laboratory in deep cooling. 08:28.300 --> 08:32.560 So far, the thunderstorm hunters have been able to measure the 08:32.560 --> 08:34.880 turbulent thunderstorm in the middle of it three times. 08:34.880 --> 08:38.940 The first result was that the tracks are very complex. 08:39.200 --> 08:42.900 It's not like this balloon or probe is simply sucked in. 08:42.920 --> 08:46.580 Instead, we see spiral-shaped tracks before they reach the upwind. 08:46.580 --> 08:50.000 Then in the upwind we also see strong turbulence. 08:50.180 --> 08:53.640 We had one thunderstorm cell, which was a very weak supercell. 08:53.760 --> 08:57.500 We measured a vertical velocity of 36 m per second. 08:58.340 --> 09:00.500 That's almost 130 km per hour. 09:00.700 --> 09:03.320 So already extremely strong vertical velocity. 09:03.480 --> 09:04.900 It surprised us how high it is. 09:05.400 --> 09:07.400 We have a total of three cases where we were successful. 09:08.180 --> 09:10.560 Of course, we need many more cases for the statistics. 09:10.760 --> 09:14.220 The main area of ​​operation for thunderstorm hunting is the Swabian 09:14.220 --> 09:15.860 Alps and the Neckar Valley. 09:16.300 --> 09:19.120 This is where most supercell thunderstorms occur. 09:19.560 --> 09:22.360 This is also where the risk of hail is highest. 09:22.680 --> 09:25.220 The hypothesis is that we see this in model simulations. 09:25.540 --> 09:29.640 We usually have a stream from the southwest during thunderstorms. 09:29.840 --> 09:32.800 Then we have the Rhine Valley, where the stream is first channeled, 09:32.900 --> 09:36.680 runs parallel to the mountains, Bogesen and Schwarzwald. 09:36.800 --> 09:40.840 Then the Schwarzwald flows around on the north and south sides. 09:40.840 --> 09:44.700 By the way, this ensures that we have a divergent groundwater flow 09:44.700 --> 09:46.980 around Karlsruhe and the Karlsruhe region. 09:47.280 --> 09:49.160 Divergence on the ground always means a descent. 09:50.000 --> 09:53.140 This could be one of the reasons why there are relatively few 09:53.140 --> 09:53.660 thunderstorms here. 09:53.740 --> 09:57.860 In the Neckar Valley, on the contrary, we have the convergence. 09:58.120 --> 10:02.980 The stream runs around the southern Schwarzwald, on the other hand, it 10:02.980 --> 10:04.020 runs around the northern Schwarzwald. 10:04.320 --> 10:07.940 They meet and form a large-scale convergence. 10:07.940 --> 10:11.700 If this is one of the hypotheses that I published in 2010, whether 10:11.700 --> 10:13.580 this is actually true, I'm not sure. 10:13.780 --> 10:17.360 What we see is that these convergence areas play a very important 10:17.360 --> 10:19.580 role, but it is probably more small-scale. 10:20.440 --> 10:25.340 We now have observations and associated evaluations over a longer 10:25.340 --> 10:29.900 period of time from, for example, radar data, that thunderstorms often 10:29.900 --> 10:31.980 flow down to this region of Feldberg. 10:32.660 --> 10:35.600 Then they continue to move into the Neckar Valley. 10:35.600 --> 10:38.580 The Neckar Valley is very humid, in fact, even more humid according to 10:38.580 --> 10:40.280 our measurements compared to the Rhine Valley. 10:40.340 --> 10:40.760 It's amazing. 10:41.640 --> 10:45.100 They can then intensify there and move over the Swabian Alps. 10:45.220 --> 10:48.100 Here we actually have the Neckar Valley, the Swabian Alps. 10:48.220 --> 10:50.960 This is the hotspot of hail events in Germany. 10:51.400 --> 10:55.140 The hail forecast of the German Weather Service is already relatively 10:55.140 --> 10:55.740 good. 10:56.140 --> 10:59.800 Nevertheless, the lift project should once again make a significant 10:59.800 --> 11:00.620 improvement. 11:01.300 --> 11:04.920 Overall, the problem is a location-precise forecast. 11:05.760 --> 11:08.580 Hail causes very small-scale extreme damage. 11:08.700 --> 11:09.680 Part of it can be avoided. 11:09.960 --> 11:12.700 Of course, if the vehicle, for example, drives into the garage, you 11:12.700 --> 11:14.820 can open up trolley shops and get rid of everything that is mobile and 11:14.820 --> 11:15.140 mobile. 11:16.200 --> 11:20.340 Only then a forecast helps you, which says, well, somewhere in the 11:20.340 --> 11:23.160 north of Baden-Württemberg, the danger of hail is relatively high, it 11:23.160 --> 11:25.240 doesn't help you, because then you would have to do it every third 11:25.240 --> 11:25.580 day. 11:25.580 --> 11:30.080 What we are talking about here are predictions of the so-called now 11:30.080 --> 11:30.560 -casting. 11:31.280 --> 11:34.220 These are predictions and warnings based on them in the next period of 11:34.220 --> 11:35.580 time, maybe two hours. 11:36.100 --> 11:38.380 This is what our research aims at. 11:38.520 --> 11:43.380 That we understand better how the trajectories actually run, how the 11:43.380 --> 11:45.900 hail growth is running, and how we can better recognize this in the 11:45.900 --> 11:47.540 data that is available operationally. 11:49.220 --> 11:52.700 In the form of radar data now, conventional radar data, but also dual 11:52.700 --> 11:56.620 -polarization radars, in combination with the convective environmental 11:56.620 --> 11:57.040 conditions. 11:57.120 --> 12:01.760 And that is the main goal, to understand better how the hail growth is 12:01.760 --> 12:05.940 running, which signatures we can observe, in order to ultimately 12:05.940 --> 12:10.300 improve this prediction and the warnings derived from it on a very 12:10.300 --> 12:12.140 short time scale, up to two hours. 12:13.520 --> 12:18.240 And the size of the expected hail grains should also be predicted. 12:18.920 --> 12:21.580 From the radar we can already say exactly, there is a thunderstorm 12:21.580 --> 12:24.520 cell, and it is now likely to move into the region. 12:24.760 --> 12:27.360 We have already come a long way with supercells. 12:27.420 --> 12:31.220 For example, cell division occurs, usually the left one dies, and the 12:31.220 --> 12:32.220 right one intensifies. 12:32.660 --> 12:34.520 This has something to do with vertical pressure disturbances. 12:35.100 --> 12:39.280 This means that these supercells can be extrapolated very well in the 12:39.280 --> 12:40.600 future, if you have the observation. 12:41.580 --> 12:45.100 But if you can still say with a relatively high probability, okay, we 12:45.100 --> 12:50.760 now have a hail of maybe 5 cm, of maybe 3 cm, then you can actually 12:50.760 --> 12:52.240 give out pretty accurate warnings. 12:52.900 --> 12:56.100 So not only on a land area, but actually also on a local level. 12:56.640 --> 13:00.560 This is of course for sensitive areas, for energy supply. 13:01.340 --> 13:05.480 If you think of the solar fields, some of the solar cells are designed 13:05.480 --> 13:06.580 to be tiltable. 13:06.940 --> 13:10.220 This means that if a hail storm arrives, then you also have the chance 13:10.220 --> 13:14.160 to tilt the solar cells out of the main wind direction, in order to 13:14.160 --> 13:15.520 ultimately avoid damage. 13:15.520 --> 13:22.320 So-called hailstones flyers try to fight the hail by emitting silver 13:22.320 --> 13:22.960 iodide in thunderstorms. 13:23.940 --> 13:25.520 Silver iodide is an INP, i.e. 13:25.540 --> 13:27.000 an ice nucleating particle. 13:27.220 --> 13:30.380 The reason why hail can get so big is that there are too few of these 13:30.380 --> 13:33.700 INPs, and the idea is just to make sure that there are more. 13:34.040 --> 13:36.960 And then there may already be hail, but the hail remains smaller 13:36.960 --> 13:36.960 overall. 13:37.180 --> 13:39.960 And if hail is relatively small, it also has a good chance to melt 13:39.960 --> 13:40.960 before it reaches the ground. 13:40.960 --> 13:46.240 Worldwide, agriculture mainly invests in hailstone prevention through 13:46.240 --> 13:47.180 silver iodide. 13:47.700 --> 13:51.140 For hailstone experts, this is a dubious investment. 13:52.020 --> 13:55.580 So far, there is no scientific evidence of its effectiveness. 13:56.040 --> 14:00.120 There is no publication on this, at least here in Germany, which 14:00.120 --> 14:00.760 proves this. 14:00.920 --> 14:04.620 There is no scientific publication in the Fachjournal, which has been 14:04.620 --> 14:05.360 peer -reviewed several times. 14:06.420 --> 14:08.020 Nevertheless, vaccination is being done. 14:08.020 --> 14:10.820 My point is, however, we don't see it in our data. 14:11.240 --> 14:15.040 We know which regions are being vaccinated, and we don't see it in our 14:15.040 --> 14:15.580 radar data. 14:16.280 --> 14:19.960 We don't see it in the damage data from insurances we work with. 14:20.120 --> 14:21.480 We don't see it in the lightning data. 14:22.340 --> 14:25.460 In the data sets over a longer period of time, we just don't see that 14:25.460 --> 14:26.040 it has any effect. 14:26.940 --> 14:30.200 I don't want to rule out the fact that in some events it actually 14:30.200 --> 14:33.120 causes the hail to get smaller. 14:33.280 --> 14:34.320 You can't rule that out either. 14:34.320 --> 14:39.060 But the question is, in other cases, could it also cause the hail to 14:39.060 --> 14:39.380 get bigger? 14:40.080 --> 14:41.040 It would be possible, yes. 14:41.640 --> 14:43.900 And in the model, of course, you can do exactly that. 14:44.140 --> 14:46.200 You can change the concentration of different aerosols. 14:48.040 --> 14:49.060 What do we see in the model? 14:49.480 --> 14:52.500 Extremely complex results, but we don't see if you change the INP 14:52.500 --> 14:54.660 concentration, that the hail gets smaller. 14:55.020 --> 14:56.600 It's not a systematic effect. 14:57.100 --> 14:58.140 And that's what it's all about. 14:58.760 --> 15:02.920 The interplay of the dynamic and thermodynamic processes is extremely 15:02.920 --> 15:04.100 complex in such a thunderstorm. 15:04.380 --> 15:05.220 It's not linear. 15:05.680 --> 15:08.900 So it's not like we say, okay, let's just take more from the aerosols 15:08.900 --> 15:09.840 and everything changes. 15:10.220 --> 15:11.200 No, that doesn't work. 15:11.400 --> 15:14.860 The thunderstorm is so dynamic and so high-grade, not linear. 15:15.080 --> 15:17.700 And that's exactly what we see in our results.