The idea is actually that the satellite images or the active sensor data provide a basis for analyzing a flight area very precisely, and that you then input into a variometer or a map what constitutes a hazard, such as lee effects or the aforementioned thermals.
Lu-Glidz, the Lu-Glidz podcast.
Stories from the cosmos of paragliding. Today
with David
Holzhacker and
Lucian Haas at the microphone.
This 21st episode of Podz-Glidz is a bit of an outlier. Until now, I've always interviewed people for the podcast who have, among other things, had special experiences, long-term work, travel destinations, or other inspiring projects involving the paraglider. But today is about a look into the future, a vision. It is the vision of David Holzhacker. David
is 28 and works as a geographer at the Julius-Maximilians-Universität Würzburg. And of course, he is a paraglider, even if only for two years. But this hobby has captured him so much that many of his thoughts now revolve around how he could use his knowledge in geography to the benefit of paragliding. David's specialty in geography is remote sensing. It's about the use of satellites, aerial, and radar images in high resolution. To learn more about the terrestrial structures we live on and over which we also fly. According to David's assessment, remote sensing methods and data could offer us pilots much better possibilities for assessing and analyzing flight areas in the future than is already possible today with known tools like Google Earth.
In
In this episode of Podz-Glidz, David Holzhacker shares his vision. Among other things, he explains how remote sensing works and what kind of data is already being generated from it today. He shows how this data is available on the internet. Additionally, he provides insights into how the systematic use of such information could change our approach to flying in the future.
David, you're a geographer and you deal with the use of satellite imagery for a wide variety of applications. And since you're also an enthusiastic paraglider, it's natural to think about how we could use satellite images meaningfully for our hobby as well. What does the view from space offer me as a pilot that I can't see from the ground? What does the view from space offer me as a pilot that I can't see from the ground? What can I discern with satellites?
Well, from space or using remote sensing—maybe I'll say remote sensing first, as it represents all measurement methods that work from a distance, like satellite imagery or aerial photography. The possibility here is that I can examine large areas from a great distance in a completely different way, and see things differently than I might perceive them on the ground.
And what does that give me now as a paraglider? I fly in my area—say, here, this is my flight area and so on. I take off into the air and look down from above. What kind of information could I get from such satellite images?
Satellite images, on the one hand, give me the opportunity to get to know the area on a large scale first. Like, what kind of terrain do I have in front of me? What's the relief like? What vegetation might be predominant? Do I have a high proportion of rock or a high proportion of forest? Where are the possible landing fields? And where are the possible out-landing options for cross-country flights? I get all this information from satellite images in great detail.
What kind of satellite imagery are we actually talking about here? I assume you don't mean the usual images we see from weather satellites. So...
There are different approaches in remote sensing. Let me give you a bit of background on that. On one hand, you have active sensors, and on the other, you have passive sensors. Active sensors are, for example, the classic radar. So, I send out radio waves, and based on the reflection of those waves, I can measure distance or get other information about the terrain. And then we have a significantly larger number of passive sensors. These are the classic satellites orbiting the Earth that capture the Earth in various ways across different wavelengths—that is to say, light spectra.
Those are primarily the types of images I work with.
But these are the kind of images where you'd say, as a geographer and a researcher, you deal with them. To what extent could I, as an average person, even get access to them? Many of these are probably special satellites that just beam down their data. Can I find those completely freely on the internet?
Well, the most well-known one is obviously Google Earth as a satellite. Or rather, Google Earth provides various images. It's not a direct satellite; it's various aerial shots, or various satellite images that are processed and have been processed. So everyone knows that, you can search for your own house on Google Earth. And beyond that, of course, there are many more satellite images that you can find through various sources, like for example Google, which has developed a special Earth Engine for this that gives everyone the possibility to access a wide variety of satellite images. And
is that one also freely available?
Exactly, those are freely available images. Of course, there are images that aren't completely free, but the vast majority of satellite imagery is freely available nowadays.
And if I say as a pilot, I'm going into a new flight area that I don't know yet. What would be sensible from your perspective? What kind of satellite images should I look at to get certain information about it?
What's really exciting, of course, are especially radar images on the one hand, because they tell me a lot about the slope gradient—which way a slope is tilted or, of course, how steeply the terrain drops off. Alternatively, I also have these so-called multispectral images available, where I can see very clearly what the forest cover is like. Do I have potential thermals? How good are my landing options? Or do I have any landing options at all? And all this information in combination gives me a very good overview of the flight area.
Now you just said the word multispectral image. What am I supposed to imagine by that? What is a multispectrum?
A multispectrum is an expanded perspective, you could say. Roughly speaking, the human eye perceives a certain wavelength. The sun emits wavelengths in the form of electromagnetic waves. These are reflected by the respective objects, and the eye then creates an image. But of course, our eyes can only see in certain colors. Blue, yellow, red—those are the familiar ones. Beyond that, everyone knows the ultraviolet range. So we're talking about wavelengths around 400 nanometers and below. That's the range responsible for sunburns, for example. And beyond that, at the end of the red spectrum, past 700 nanometers, we have the infrared range.
And all these wavelengths in combination are then multispectral satellite images. So I have significantly more colors that I'm displaying, which I can't perceive at all in a simple photo or with my own eyes.
But when I want to look at them—as I said, I can't see infrared or anything like that—then I only get them displayed as a kind of false-color image.
Exactly. So I just color the area I can't see with my own eyes in a different color. For example, I might say that the infrared range should be represented in red.
Now you just said it could be interesting to look at these kinds of images to see, for example, where the forest is and what else. And you also mentioned the word where thermals might be. How can I identify thermal sources based on satellite images?
Well, thermals form in certain places, as we're all aware. We have certain slope gradients. We have dark areas that heat up particularly well. And I look for these sources in these multispectral images very specifically by, for example, getting information on how green a forest is. Does it perhaps have a high water content? Is there a certain amount of evaporation that has to take place first before the thermals can form? I get all this information from these multispectral images at an incredibly detailed scale.
What do you mean by "incredible"? To what zoom level can I imagine? Is it just down to one meter, 10 meters, 50, 100 meters? What kind of scale is that?
So we have the option where classic satellite images capture between 30 meters per pixel to 10 meters per pixel. Newer satellites also already capture at three meters or one meter per pixel. That means a one-meter field would basically be one pixel on my image. The much more exciting thing about it is actually that through various spectral information and indices that I apply specifically to vegetation, I get very detailed information about the degree of vegetation, the exposure of the slope, the slope's inclination, and the reflectivity or also the absorptivity towards certain wavelengths, and with that, of course, the possibility and the chances of thermals.
That means, for example, I could identify that this slope absorbs particularly much heat at a specific spot, or another one that reflects much more of certain infrared waves or others, and I can then see where, for instance, it heats up a bit more and where a bit less. Exactly right. But this is a satellite image, which was created at some point in time. It changes depending on the time of day, the season, and so on. How can something like that actually help me in practice?
So the satellites are constantly taking images of the Earth, and certain parameters don't change very much at all. Especially things like forests or rocks—they're always in the same place. My idea behind this approach is a combination of different types of information. Of course, if I have a perfect slope for thermals and the sun isn't shining, it won't help me regardless. But the combination of weather, satellite imagery, and of course active sensors gives me the possibility to get an incredibly detailed representation of the respective flight terrain.
That
That all sounds interesting to me for now, but at the same time, it's incredibly complex. You have to understand so much. You have to understand a lot about how to read these false-color images in the first place, and what information I can actually get out of them. For me as a pilot, it would really only be interesting if someone already aggregated that information for me in a way that's usable. From your perspective, is there anything already on the market—apps, internet programs, or something like that—where I could actually say, "I can get real value out of this today"?
Of course, there are some apps that provide certain information. For instance, things like slope exposure or simpler satellite imagery, or even apps combined with weather data, already exist. However, my idea is also to develop an app or a methodology that could then be integrated into, for example, variometers or into special terrain maps that provide exactly this information to pilots who have never flown in that terrain before.
What would something like that look like from your perspective? I mean, what do you imagine? This is like a vision you have right now. Yes,
Vision actually hits the nail on the head. The idea is that the satellite images serve as the basis. So, the pilot shouldn't actually have to do the interpretation directly in the air, which isn't possible anyway. You're busy with other things while flying. The idea is actually that the satellite images or the active sensor data represent a basis with which you can analyze a flight area very precisely, and then you feed into a variometer or a map what constitutes a hazard source, such as lee effects or the mentioned thermal effects. And the possibilities there are a combination of the most diverse data, which is then very well processed and made available to all pilots.
We already know this, for example, from some variometers that include no-fly zones in their maps.
That means you'd say, I'll include thermals zones, I'll include lee zones and stuff like that. But that's also constantly changing. I mean, the wind can blow from the north or the south.
Simply put, from that perspective, the lee can be on one side or the other of the slope. I can't necessarily tell that, or how the wind is currently blowing, just by looking at a satellite image.
No, exactly. You can't tell from that. However, a leech requires a certain slope or a certain hillside. And that can then be simulated with the wind. That means I have my base data, my satellite data, my slope, my vegetation, maybe even my terrain classification, and then I simulate what happens with lee waves when the wind comes from the north, when the wind comes from the south. That's sort of a type of model information.
But if I see that displayed on my vario, I as a pilot still have to decide which model information I'm trusting. Does it actually match my reality right now?
Exactly, right. That's obviously the important information that I have, which I then have to assess during the flight. I haven't been flying for an infinite amount of time myself either; I've been flying since 2017.
And although I've flown a lot in these last two years, especially at the beginning, I always felt a bit uneasy when I went to new flying areas alone or with friends who hadn't been flying for very long either. And that's where this idea came from: that you shouldn't only get this information actively while flying, but that this information should be made available in detail for flying terrain beforehand. So, especially when you think about active satellites, satellite sensors, laser sensors—I naturally have the possibility to create incredibly good 3D models of flying terrain that have a level of detail in the millimeter range.
And that's exactly what I sometimes wished for when I went to new flying areas. I mean, of course I looked at Google Maps and Paragliding Map and whatever other sources of information there are. But I was always missing that sort of 3D model that I could look at at home, that I could rotate, where I could zoom in on the exact details to see what theoretical conditions I have to work with here.
You mentioned Google Earth earlier, and that actually works quite well. I do use Google Earth sometimes; I'll say where I'm going, then check it out in Google Earth, rotate that quasi 3D model, and look at a mountain from different angles. However, I constantly have to experience that what you see live on-site with your own eyes—from different angles and everything else—often looks different than it did in Google Earth. Isn't that the danger, even if you say you rely too much on this remote sensing technology? You have to say, well, it looks different than how I see it, and I might get the wrong impression of the terrain.
Of course, there's always the risk that I rely too much on technology, saying I'm only looking at terrain data from remote sensing and then flying that terrain. The possibilities of, for example, using laser data that is so high-resolution that I can see almost every pothole at the landing field or the launch site, offer a different level of detail than Google Earth. I think Google Earth is a good way to get a rough overview of a terrain or to get to know it roughly from a distance. I think that with a targeted, methodical application, specifically for paragliding, these data could be significantly improved and made available, especially to inexperienced or new pilots.
What would that look like in practice? Do you imagine paragliding clubs saying, "This is our territory, let's have a plane with a laser surveying device fly over it now, they should provide us with a proper three-dimensional model of it, and then we'll make it available to our pilots," or how is that supposed to work?
Exactly, so that would be the ideal way, which of course is always associated with a lot of funding, with a lot of financial effort. In my own work, I've sometimes done the whole thing with drones, and not necessarily with laser sensors, but with completely normal camera sensors, which then allow for a very detailed view of certain terrain through these photogrammetric representations—meaning 3D modeling.
Airspace holders or clubs would be an interest group that could obtain this data, but you could also market or provide the data via a website and coordinate specifically with certain flight areas.
What do you mean by "tailoring"? I mean, what would be offered there?
So from my experience here in Allgäu, there were always flight areas that were flown a lot by beginners and, of course, by me too, and flight areas that I always steered clear of at the beginning, not because the flight area itself might not have been that difficult to fly, but because, for example, the landing field was too small for me at the start, or because a slope was so steep that I simply didn't dare to fly there. And I think, especially for beginners, this kind of representation is incredibly helpful, because before I actually look at the terrain on-site, I can get a sense of whether I even want to fly there, and whether I feel confident enough with my level of experience to fly there at all.
And for me, it was often the case that I'd be in a flight area that sounded great, and then I'd stand at the landing field and think, with today's wind conditions and this landing field, I don't dare to land here, despite my ass. That naturally changed over time, but especially at the beginning, I could have saved myself the drive, for example.
Can you give a concrete example of when something like that happened, and how could you use the tools available today, whether it's Google Earth or something else, to maybe actually avoid that, so it doesn't happen and you don't fly into such terrain for nothing? I mean...
For me, a landing field like the one here in Allgäu at Hochgrat, where I wanted to fly right at the beginning but didn't actually fly, is a case in point. The landing field is slightly sloping, meaning it has a slight incline. On top of that, it's surrounded by trees on two sides. There are also some rocks on this landing field, and there's a relatively large ditch at the front that you definitely shouldn't fly into. That means you have to plan and have a very good grip on the final approach as well as the headwind and crosswind approaches to land there very safely. Of course, for many local pilots, that's no problem at all. For me as a beginner, it was already a problem back then.
And I might not have driven to this landing field or this flight area independently if I had known what the landing fields were like back then. And what the conditions were like. Nowadays, that's no longer a problem. But especially at the beginning, fresh with my private pilot license, it was just a bit too challenging.
Now, if such data is prepared, in whatever form, you would also need experience with the processing. So that you know, if the representation is like this and that, it's a landing field I can still handle. And if the representation is like this and that, you'd say, that's a landing field I might not dare to use anymore, and so on. So there would have to be a learning effect, an experience factor, first of all. Do you really think that these satellite images help that much?
I think a combination of satellite imagery and laser data could be very helpful there. While Google Earth is already very detailed, of course, I also looked at the landing field in Google Earth back then and saw, well, we have trees, we have stones—you can see those less—and how deep a ditch is, you can't really estimate that on Google Earth either. But with a 3D model, which are very detailed by now, you can estimate that much better. I've created these 3D models based on drone data as a trial for some meadows or some slopes, just for myself.
And I was surprised myself by how detailed the representation is and how well it matches reality.
If you were to create a 3D model of an airfield using a drone, as you call it, how much effort is that? How long do you have to fly back and forth with the drone? How many thousands of images do you have to take so that you can say, photogrammetrically, you can derive a somewhat usable model from it in the end?
Yes, that is naturally... a certain amount of work. As you correctly said, you fly the drone there, and then it depends a bit on what kind of camera I have. Is it more of a zoom lens with a detailed view, or am I capturing a larger area? The second question is whether I actually need the entire terrain or if, in some cases, landing and takeoff spots might suffice? Then, of course, it's significantly less work. Overall, you do have to fly the entire terrain with the drone and take a picture of the ground every few meters to represent it in these photogrammetric 3D models.
So, I think for an entire flight area, you'd end up with a few hundred to a thousand images.
And your idea would be that, ideally, you'd have a database where 3D models of many different take-off and landing sites are stored, which you could then work with.
Exactly, that's the basic idea—that every pilot can access it without having to dive deep into the analysis themselves, but can still see whether a piece of terrain is very suitable or less suitable for flying.
Are you aware of any initiatives that are already working in this direction? I mean...
I acquired my remote sensing knowledge during an internship at the German Aerospace Center, and I also regularly consult with the center regarding methodology and approach. As far as it being designed for paragliding, I don't know of anything yet. However, there is already a lot of interest in processing this data as well.
What do you mean by "also these data"?
So, that you combine this data from remote sensing with paragliding. The data itself is just raw material for all kinds of questions, whether they are of a biological nature or whether they are potential hazards. The idea is, of course, always to say: what question do I have? In my case, it's paragliding. And what can these remote sensing data provide for that? And how can I also collect some of this data myself, as described, using drones?
Do you have plans to continue working in this direction yourself?
Yes, well, it's actually a side project of mine right now, where I'm saying I want to start creating a database, at least for the flight sites, initially where I naturally love to fly, which is a lot in the Allgäu, and to definitely make these images and these models available to see how pilots actually receive it. Is there interest in it? And I'm also of course interested in the feedback from pilots who then tell me, "I would have wanted even more," or "I would have maybe wanted more detail." It's supposed to be a way of observing that, at the beginning, we become a platform with which I'd also like to work a lot with feedback.
If I were to come onto this platform now—that's still your vision, but if I were to come onto this platform now—what exactly would I find? I mean, what kind of information would I be able to get from it?
First of all, you'd primarily get a very detailed 3D model of the terrain that you can zoom into and rotate, where you can see where I have slope exposures, where I have forest, where I have rocks, where I have possible lee effects—even without knowing the wind exactly—so where do I have them, or also where do I have ridge lines for the thermals? That's one approach or one way of looking at this model, of course. Another consideration would be to overlay this model with these multispectral images. That means, where do I have better thermals, which thermal sources based on the information? Where do I actually have lee effects under which wind conditions?
That's my vision and the idea behind the whole thing.
It's still just the vision, you say. When can this vision at least become a reality on a small scale as an initial example?
Yes, I'm actually hoping to create a first model of a flight area in 2020. Or that the 3D model is definitely finished by then. And that I can already apply the first multispectral images to this model. To then see how the whole thing comes across and what the feedback is from
Pilots? Do you think that at some point in the future there could be a kind of automatic evaluation? That maybe a type of artificial intelligence could be used that then says, "Look, you're looking at this 3D model, you're looking at the multispectral images, you're looking at the multispectral recordings, and in the end, filter out for me so I can say, show me all potential thermals release points," and then, boom, they appear on my screen marked accordingly?
Yes, that would obviously be the absolute ideal path and the dream scenario, so to speak, to have a completely automated process. Of course, you can't automate the whole thing, because the images still have to be collected first, or more importantly, the recordings of the terrain have to be created first. But exactly as you say, with an artificial intelligence that also learns over time, an interface with the pilots is also important—they tell me, "This thermal was predicted by the AI, but it wasn't there today, it was somewhere else." So it's also important that this whole project is created in collaboration with pilots and that there's some kind of feedback that tells me, "I'm doing it right and my methodical evaluations actually work in practice too."
But don't you also think that when you use so much technology and have so much displayed in advance and whatnot, doesn't it also take away a bit of the fascination of flying, where you really have to adapt to new things, to the conditions that exist at that very moment, and where that's actually the challenge of flying—that you actually have to expose yourself to the unknown in that moment?
Yes, that's actually a very good question. The pilot in me says it does take away the fascination of the unknown to some extent. The geographer, of course, says, I want to know everything I can. But I think for me, it's less about minimizing or completely removing that thrill of the unknown or the residual risk. It's more about developing an additional safety standard that provides me with information that can help me, but isn't unnecessary knowledge about the aircraft and the conditions.
So I think technology can help, but it can by no means replace the pilot.
Okay, that's your vision as a geographer overall, based on what you've just described. Is there any other insight from your geography studies, or from the geographical knowledge you've gained, that helps you out today when you're flying?
Yes, consciously, maybe always a little unconsciously. Well, of course, you always look at the world a bit differently, let's say, when I'm sitting in my harness as a geographer. I naturally see many slopes or many flight areas from a geographical perspective as well. So, what kind of features do I have here, how did they come about, I mean, how was this whole landscape shaped? And of course, through geography, which naturally also covers part of meteorology, I have a lot of knowledge where I say, yeah, I'm familiar with the weather events here, for example, or I know how I can approach it when the sun is shining like this and when there are dangers from the Föhn.
So, even in the sense of it, I didn't find this meteorological perspective that difficult. How
did you ever actually get around to flying? That
It actually goes back quite a while. So my neighbor, she flew a very, very long paraglider back then and I always thought it was totally great whenever I read the DHV booklet at her place. And yeah, at the time I was, I think, 14, something like that. And yeah, paragliding was in the far distance, so to speak. But I never completely lost sight of it. And after my studies, I think I saw a film of two guys who flew in Mexico. They hiked up there—I don't even remember which mountain it was—and flew down. And that fascinated me so much that I said, and now the point has come, now I'm signing up for the license.
And then I signed up and haven't stopped since. Was that by any chance
The movie The Fletchlings? Yes, I think so. Yes, I think, I think that's what it was called. They're two American mountaineers who actually come from mountaineering and then started learning paragliding. And they then, I believe, climbed Popocatepetl or something, a large volcano in Mexico, to then fly down from there. Which, however, was a relatively chaotic project, if I remember correctly.
I also think it was chaotic. I believe the film was shown at some point at the Banff Mountain Festival. I saw it then. I had absolutely no idea about flying back then. To me, it just looked incredibly cool, how they climbed up there. And then saw the world from above like that. In hindsight, if I were to watch the film again, probably a lot of it is actually much more chaotic than I perceived it at the time. But that was actually the turning point for me, where I said, "Okay, now I'm going to get my license."
Then you got your license. And since then, the more you've flown, the more you've probably gained that vision. How can I combine my geographical knowledge and those satellite images and remote sensing and so on, which you also deal with, with the paraglider world? David, thank you for your presentation. That's an interesting look into what might come in the future. And I'm curious to see what you'll do with your website, where hopefully next year the project will be visible, so people can get a better idea of where the journey might be heading.
Yes, thank you very much for the interview.
You're very welcome.
That was David Holzhacker in conversation with Lucian Haas. If you're interested in more tools and methods from the TV show, I've put together some relevant links in the show notes for this episode of Podz-Glidz on Loglites. Anyone who wants to get in touch with David himself can best do so via email. The address is david.holzhacker at gmail.com. Finally, a quick announcement for my own sake. As always, you can support my work on Podz-Glidz and the Loglites blog as a patron. I'll leave it up to you how much you want to give as a contribution. As a non-binding guide, I recommend 1 Euro per podcast episode and 2 Euro per reading month on Loglites. Payments are easily possible via PayPal or bank transfer.
You can find the relevant data on the Loglites website, specifically on the Support page. Thanks for listening and for your support. We'll continue with new episodes of Podz-Glidz in 2020. Until then, stay safe. Ciao and happy landings.