Investing in Regenerative Agriculture and Food
Investing in Regenerative Agriculture and Food podcast features the pioneers in the regenerative food and agriculture space to learn more on how to put our money to work to regenerate soil, people, local communities and ecosystems while making an appropriate and fair return. Hosted by Koen van Seijen.
Investing in Regenerative Agriculture and Food
355 Tom Hengl - We should reward the stewards of the land like we celebrate Olympic champions
A long-overdue check-in conversation with Tom Hengl, director at OpenGeoHub, one of the leading scientists in earth observation and remote sensing—one of the most cited in his field, belonging to the top 0.1% (based on Clarivate Highly Cited Researchers). We discuss the significant changes in the world of remote sensing, satellites, and the hype surrounding AI, machine learning, and large language models over the past three years. While the hype has brought some interesting advancements, it also distracts people from the real work that needs to be done.
We delve into the AI4SoilHealth European project we are part of, discussing how we can already monitor and observe most places on Earth from the sky at a resolution of 30 by 30 meters. Importantly, we can now look back nearly 25 years for almost all locations in Europe and analyse changes on a field-by-field basis. While we might not know the individual farmers, we can identify their fields, and we can train models to make predictions and provide actionable, relevant advice.
We explore the idea of celebrating farmers and land stewards who have successfully regenerated their plots of land over the past decades. But how do we shift a culture that celebrates sports over regenerative farming? Finally, we touch on the challenges holding back some of this work, including the need for reliable and affordable in situ in-field soil health analysis.
More about this episode on https://investinginregenerativeagriculture.com/tom-hengl.
This podcast is part of the AI 4 Soil Health project which aims to help farmers and policy makers by providing new tools powered by AI to monitor and predict soil health across Europe. For more information visit ai4soilhealth.eu.
Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or European Research Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.
This work has received funding from UK Research and Innovation (UKRI) under the UK government’s Horizon Europe funding guarantee [grant numbers 10053484, 1005216, 1006329].
This work has received funding from the Swiss State Secretariat for Education, Research and Innovation (SERI).
==========================
In Investing in Regenerative Agriculture and Food podcast show we talk to the pioneers in the regenerative food and agriculture space to learn more on how to put our money to work to regenerate soil, people, local communities and ecosystems while making an appropriate and fair return. Hosted by Koen van Seijen.
==========================
👩🏻💻 VISIT OUR WEBSITE https://investinginregenerativeagriculture.com/
📚 JOIN OUR VIDEO COURSE: https://investinginregenerativeagriculture.com/c
Thoughts? Ideas? Questions? Send us a message!
Find out more about our Generation-Re investment syndicate:
https://gen-re.land/
https://regenerativo.org/en/laris/
Feedback, ideas, suggestions?
- Twitter @KoenvanSeijen
- Get in touch www.investinginregenerativeagriculture.com
Join our newsletter on www.eepurl.com/cxU33P!
Support the show
Thanks for listening and sharing!
We give awards to sports people. Somebody goes, wins the Olympics medals and it's all in the news, the pictures and their girlfriend. And what about the champions of land restoration? We should also find these people. We should celebrate them. We have a scientific proof that they're champions, not because there's some political party. But we say no, we are independently looking. And this is the top list.
Speaker 2:A long overdue check-in conversation with one of the leading scientists when it comes to Earth observation and remote sensing, one of the most cited in his field. He belongs to the 0.01%. We talk about how much has changed in the almost three years in the world of remote sensing, satellites and all the hype around AI and machine learning and large language models, llms. It's a lot of hype and some interesting things, but it also distracts people from the real work which needs to be done. We discuss AI for SoHealth, the European project we are in together, and how we can already monitor and observe most places on a 30 by 30 meter scale from the sky and, importantly, we can now go back on most places in Europe 25 years and see what happened on a field by field basis. We might not know the farmer, but we know his or her fields and we can train models to look forward and start give actual, relevant advice. We explore how we actually can and should celebrate the farmers and land stewards who have regenerated their plots of land the most over the last decades, objectively. But how do we change a culture which celebrates sports people instead of regenerative farmers? And what's holding some of this work back? Reliable, cheap, in-situ, in-field soil health analysis. This is the investing in regenerative agriculture and food podcast. Investing as if the planet mattered, where we talk to the pioneers in Regenerative Agriculture and Food podcast. Investing as if the planet mattered, where we talk to the pioneers in the regenerative food and agriculture space to learn more on how to put our money to work to regenerate soil, people, local communities and ecosystems, while making an appropriate and fair return. Why my focus on soil and regeneration? Because so many of the pressing issues we face today have their roots in how we treat our land and sea, grow our food and what we eat, wear and consume, and it's time that we, as investors big and small and consumers, start paying much more attention to the dirt slash, soil underneath our feet. This podcast is part of the AI for Soil Health project, which aims to help farmers and policymakers by providing new tools powered by AI to monitor and predict soil health across Europe. For more information, visit AI4SoilHealtheu or find the link below.
Speaker 2:Welcome to another episode. Today. We welcome back Tom, the founder of Envirometrix and OpenGeoHub. Welcome, tom, hi. So we had you on, together with Ish, obviously your partner in crime in life and in business, about I checked in March 2022.
Speaker 2:But, of course, many things have changed in general in the world, but I think in the world of machine learning, ai, large language models, etc. We cannot point at a few years where more, where less has changed. I think, like this is, or where more has changed, this has been, it felt like probably a century worth of development squeezed into three years. So I'm very much looking forward to check in what you see. You're one of the top scientists in this space and focused on soil health and focus on regeneration, and we've been working together in a European funded project AI for Soil Health, so and we're about halfway in so I'm also looking forward to check in with you on that.
Speaker 2:But let's start first with where do we find ourselves now? We're at the beginning of 2025. 24 was a crazy year for anything to do with machine learning and AI. What, where do you feel we are Like? What do you? We're now at the beginning of a new year. What do you? What excites you, let's say, about all this news over news and fundraising and all kinds of models that can do all kinds of stuff? But what do you see? As somebody who's in the inside, who looks in the machine rooms of these things. What excites you at the moment on this side?
Speaker 1:Okay, well, just first to say you know, things always change with the people. I mean, last, from industrial evolution, you know everything is going exponentially. Let's say, and so this thing, that what happened with chat, gpt, which you know didn't exist more or less last time we spoke, and then it just like took off and became one of the fastest, most dynamic development in the history of digital technology. But it's also hype. Let's put it also honestly. It's a hype. It's not that the computer started talking or thinking, they became a sentient or something conscious. It is larger hype, conscience. So it is larger hype. I mean it is quite impressive that today you can have some, you know, big models fitted big LLMs, fitted with a lot of data. They could write documents or abstracts or letters that are credible. You know that you cannot, they're not perfect but really good.
Speaker 1:You cannot tell anymore whether there was a human or not. And we did, by the way. Last year. We did a nice exercise. We called people and we will display a slide. It was about 80 people and then people have to write whether it's a it was an artwork, artwork, and whether it's AI generated or generated by a person you know, and I think the accuracy was 45%, so so yeah people.
Speaker 2:It's better to flip a coin.
Speaker 1:It's kind of.
Speaker 2:It's kind of a bit of a test you know um but like in your work, like in the remote sensing space, like making sense of imagery that comes from. Yeah, that's just.
Speaker 1:I'm just talking now about these llms you know this, large language models and generative, called generative ai. So that's the, the ai you use to produce, you know, text, images and things, um, and and behind it you know it's kind of machine learning, neural network models. You know, um, and so I'm just talking about that. So there's been a bit of a hype. Uh, I'm not too too much impressed with it, you know I I still don't use it, uh for like complex things, for solving problems, uh for like checking some statistical modeling. I mean, for me it's just like a very advanced search and aggregation tool at the moment, um, however, you know it's, uh, it's happening and the next version will be better and and eventually it might be able to also propose new solutions and it might replace a lot of jobs. You know we're talking like maybe 60% of jobs in 10 years, so a lot of jobs, especially all the service jobs they even mentioned, I think, like doctors and lawyers. They might be also in this danger zone.
Speaker 1:Danger, danger zone, yes, but still, for me it's still a hype, and I don't think for our project, for example, for Air, for Soil, health and for our work, you know, producing next-generation soil agronomy data. I don't think I'm too excited about it yet. I'm seeing some advantages and there are some developments now where because most of LLMs you know that people are familiar with, like the Google's Gemini and OpenAI, gpt and others they are mainly like they generate text or images right, so that comes out and video.
Speaker 2:Now I've seen some examples, but it's just a moving image.
Speaker 1:So that's. But what? In our field? You know, we're looking for these geospatial LLMs. So that's what you can extract from geospatial data and from these spatial temporal data cubes and that could speed up. We like it because it could. You know all the complex data. The more complex the data, the less users, you have right. The more complex the data, the less users you have right. And the more multidimensional data, scientific data, the more dimensions, the less users, the higher the complexity. And I think this geospatial technology could help reduce that complexity and make it a bit more accessible, a bit more easier for people to go into these large data pools and get something. Okay, I understand this.
Speaker 2:This is relevant for my field One example is meteorological climatic data.
Speaker 1:So let's say climatic data. Climatic data is usually about the past, the weather in the past aggregated over months or seasons or years past. You know the the weather in the past aggregated over months or seasons or years. And so somebody could go and say now if you have a like a climatic data cube, you could train the llm to get, for example, some summary. For example, say, show me all the countries that have a decrease in the rainfall over the last 30 years.
Speaker 1:That's a typical thing you will need to ask and you will need this data, you know, and you could ask me to do it and I would say, okay, I do it and I will make a code. I would say, okay, I see what you mean. So I need, for example, map of the countries, I need to run aggregation, I need to calculate the changes over years, and then I have to sort them and I have to pick up the one at the top and I say this is the country. And now llm can basically generate, because you know, llm is also used for coding. So llm will generate a code, a python, python code, r code, I don't know which will use these variables, exactly how they are in the system, and it will run very quickly. If it's a short operation, it will run a few seconds and then it will say here's the result and it can provide textual output, but it can also provide a map. Also, it can be scientific output, you know.
Speaker 1:It could be also interactive map, and that's really magical because you're basically the speed up of like. If you look, for example, 10 years ago and now or next few years, the speed up will be like really exponential. So because 10 years ago somebody will ask you that question and you will say I'll come back to you in about one week or two weeks. You know I have to, you know, make a code, test it.
Speaker 2:And now you come back with a map.
Speaker 1:And now you don't need an analyst, you don't need an expert at all. You know it's just the data and the user you understand, so there's nobody in between. That which is liberating. It's absolutely as I said. It can really speed up people using data, people making decisions with the scientific data, because we do have this problem increasingly. You know like, for example, in European Union, there's a European Commission I think it's 90 billion euros a year invested into research and it's very efficiently. This money goes and people create new data. It's very efficient. This money goes and people create new data. There's also European Space Agency has this Earth observation programs, you know, like Sentinel, copernicus, sentinel, and then they generate so much new data and it's a fantastic data.
Speaker 1:We are now in a problem that we don't have capacity anymore to manually to analyze the data. For example, we have less and less time manually to analyze the data to we, for example, we opened you up. We have less and less time just to open the maps, to look at the maps. We have increasing problem. We have like this, you know, like a 10 000 or 100 000 layers and we cannot visualize it. I mean, we have a problem too because we cannot just open every map and look, you know. So we need some ways that we can even visualize and summarize that data before we can look at things. And so we have more and more of this gap.
Speaker 1:And what AI technology is going to help? Still? People will be able to much faster get summaries, trends in the data, get information that you can use for decisions, but without any analyst in between. So you could have potentially at the same time you know, I don't know 100,000 people accessing the same data and getting outputs, and that's very exciting. That's very exciting. And also, you know the new Earth Reservation missions I don't. Also, you know the new Earth Reservation missions. I don't know if you know that, but many people are not aware of it.
Speaker 2:They probably are.
Speaker 1:There's so many things happening. This is way more now. I think it's way more dynamic because it's systematic. For example, both European Union and United States and Japan and many of these countries. They have very systematic plan to increase number of satellites, diversity of satellites, spatial resolution, revisit times. Now you have first hyperspectral satellite open data.
Speaker 2:What does that mean? Hyperspectral, what is?
Speaker 1:hyperspectral. So the Landsat is the original, you know, land monitoring program Started, I don't know, 60s, 70s, and from the Landsat 5, 6, you know it became a bit more interesting because it was high resolution and it had seven or eight bands, uh, so they felt like eight bands and uh, one bed is, for example, near infrared, you know so, but it doesn't see. In that bed, you don't see the differences within that uh part of spectra. There's no differences, it doesn't see. It kind of uh smooths out the signal, uh. So now with the hyperspectral you, you get like uh, I don't know, 800 bands, thousand beds, 1200 bands, wow.
Speaker 2:So so let's say that the resolution or the, the, you can see way more potentially, but it's also creates way more data than 10 years ago.
Speaker 1:You, you, you're like in a position to uh, basically, type of material to detect. They're testing it now still. But you know you can detect vegetation, you know species. You know not just here is vegetation, but you can say this is this species, this is this species, and so that's really amazing Not to talk about. You know geological material and soil and stuff, so you will be able to detect in much finer precision and much better accuracy, probably many things.
Speaker 2:But then what do we do? Because that will generate an infinite amount more data, even more compared to now, like what the terabytes and terabytes or petabytes, we are already in petabytes scales.
Speaker 1:So, yeah, what can you do? So then you know the part of the European Union, european Space Agency, when they make some program, you know, then they say, ok, okay, we also have to uh dedicate some budget, uh to uh test, develop applications and to see, uh test also accuracy and things, and then slowly then they test that you know it's always goes uh with iterations, uh, it goes like a couple of years. They test and they say, okay, this doesn't work, maybe we need another satellite and they can change. Uh, but it's a huge interest is in, for example, the emissions. I think sentinel 3, uh, it's uh so emissions of gases, greenhouse gases and uh also pollution. You know that's a huge interest in that and you would like to know every day basically what happens. Then Sentinel-2 is a huge interest in high resolution, 10 meter to follow biophysical indices globally, and I think you get images, I think every second week or something. So that's also really powerful.
Speaker 1:Then you have also a corporation, company which is with us in the area for solar health. We're very happy Planetcom. They used to be called Planet Labs. I think we're very excited they're with us. They have also a suite of new things services data. They're very ambitious, of course. I think they have these smaller satellites, these CubeSat satellites, so they have a higher quantity and higher quantity allows them to do much faster revisit times. So now they are aiming at daily revisit time and they're aiming at one meter resolution. They already made some global products at one meter there. They already made some global products at one meter. There is a canopy height product of the world at one meter it's.
Speaker 2:It's a mind-blowing, and canopy height means you can see how high exactly you can see the height of the trees.
Speaker 1:You can see also height of buildings and all the construction sites and things at one meter you can see individual trees you can see individual trees also. You, you have multiple pixels for one tree, you know, so you could also quantify that tree, but that's within the company.
Speaker 2:But, like, how does that work with data? Well, I think as far as I understand.
Speaker 1:I mean, I don't want to represent them. You know I don't work for them.
Speaker 1:But I have a feeling that a standard business model today it's a kind of you pay subscription. You know you will register to, for example, if you own the land or some area, if you're a farmer, I don't know, and you say, well, I would like for my farm, I would like to get from you the best data and I just need it for my farm, you know. So you will pay per square area, i't know, and you get, like you know, every week you will get, or maybe every day you get the data, and so it becomes like a, you know, like a telephone provider or like many things a weather, premium weather exactly you, you pay some subscription and and you, they basically feed you with this data and then they feed you with better data.
Speaker 1:And I don't want to say it's kind of like a Netflix, but you know what I mean. It's a new business model with this subscription. It kind of companies are becoming like governments a bit, because you know government also has a subscription, so that's the taxes you pay, right, so you don't really sell the products anymore, you kind of just register for subscription. But it is still exciting because Planet, you know they're very good at it and they can make really effective products. And these products, probably if you manage land and if you really wanted the highest quality data very fast, they could really help you manage land and if you really wanted the highest quality data very fast, they could really help you manage land. And then you will pay just this cost of that service. It will pay itself off. It will pay itself off because it helps you optimize your system. If you make a saving somewhere like 20%, you will, for sure you will pay off that service. So it's a win-win.
Speaker 2:It's a win-win, of course. How does this data lead to farmers or land stewards or forest owners or investors to make different decisions? How do we make sure because you mentioned before how we're now able, in some cases, to turn the data into something more visual, which probably is already more accessible for many people, for many people, but how do we make sure that many people not just one percent or 0.1 is even able to understand what is presented to them and, like, gives them options to okay, what should I do tomorrow to change something? Is this good, is this bad? Is this going in the right direction or not going in the right direction? Like, how do we make, give people decision-making?
Speaker 1:so, if you if you look at the like, the biggest decision, uh, services, you know like, let let's just take, for example, google Maps, right? So apparently every day, 1.5 billion people uses Google Maps and if you say this is like usually a father, if you take the family, so also, you know you're talking. Maybe half of the planet uses Google Maps every day. So what is Google Maps? Uses google maps every day. So what is google maps? Usually people use it to to find something, to to navigate in the car, to find some telephone number they have to call, because it's also integrates everything, then to check, you know, connections when they travel, trains, flights, I don't know so. So it's all kind of now integrated. So that's a really professional digital service.
Speaker 1:And now you say, well, how do we get things we do in air for? So health, how do we get it to that level? So it's a massively used, massively becomes really massive. And uh, and many, many are still not there. I think, planet from commercial ones, what I saw, they are very ambitious and as, as I said, they have their own satellites. I think they produce that's what I heard. I don't know, it's a rumor, but they produce actually more data than European Space Agency.
Speaker 1:So whether that's the same comparable, whether it's the same quality of data, that's irrelevant, but let's say, if you just look at the, you know petabytes, they claim to produce more data. So, so now let's say they produce this data and they will. They can optimize it for clients. You know they can make it. And if they make it like a netflix that you pay, um, you know most of services today. I think people are ready to pay, like you know, maybe 10, 10, 10 euros or 10 dollars a month. You know that's, that's okay, that's fine, I pay that it's not. Uh, you know, if it's less, it's better, right, um, but uh, so when it gets to that level, then it can become a massive usage, uh, and it has to be optimized for the, for the clients.
Speaker 1:I think many farmers, they're also businessmen, farmers, I mean the. There are some farmers a bit religious, let's say. I met some, but most of them they are businessmen, right, so farmers will immediately go into that if they see that they can also profit from it. You know, if you can show them, you know like, look, you can, uh, you can decrease your cost, you can optimize your harvest, you can.
Speaker 2:You can see what's happening in that field Also, if you have some damage.
Speaker 1:If you have some damage on the farm and we can provide you how much damage, where's the damage, so you can get insurance money and things like that. So if businesses convince farmers into that, then I think many will go. They will say, yeah, that's very interesting. The other thing is the policy. The policy is also going to change. You see, it's changing slowly.
Speaker 1:We have this Green New Deal for the European Union. For the European Union, there's a whole list of milestones that we would like to reach by 2030, 2040, 2050, and many would have to transition. So we are at least European Union has accepted this transition. Let's say now people vote a bit less. Unfortunately, they vote more anti-immigration and things anti anti-eu. But let's say it will still stay and so we will go to this green deal.
Speaker 1:Transition everybody has to transition. Transition means, most importantly, reduce to zero, if possible, greece house causes, switch to the alternative sources of energy heat pumps, solar energy, nuclear energy, I don't know and also then we have to recycle. We have to recycle much better materials. We have to produce more food in a smaller area. We have to minimize the losses. We have to also adjust to the climate change. So some places you'll have to change because there will be more droughts, some places will get hotter, I don't know.
Speaker 1:So all these things, you know there's a lot of work to be done and this is where these technologies, they, can play a very important role, because anything with the land, you start with the planning, and for planning you need the maps, you need the geographic or geospatial data, and then also the culture is now changing. You know, for example, if you look at some agriculture, let's say 50 years ago, even 20 years ago, if you go to a farm and if you ask, what did you do? What was in this farm last 50 years? What has been done? What have you put inside? What have you changed?
Speaker 2:What did you take out?
Speaker 1:What was your sole carbon 20 years ago? Many people don't know. Many farmers, they go from year to year, they, you know, seed and harvest and then they leave from the output, but many people don't know what is really happening and it's really too scientific. You know, many things are like still quite scientific today, like soil biology. You know, still quite scientific today, like soil biology. Um, you know, uh, this plastic, uh, this microplastic and other pollutants, people, people have no idea, you know, they don't even. Oh, wow, you can measure that. Um, and now this thing is, I think it's going to change and I think we could, within, uh, let's say, 10 years, we could go to like a total monitoring system, total monitoring and environmental monitoring. Now we do have a monitoring like we have in the cities and urban areas. They are heavily monitored. You know, you have the cameras for security, the sensors.
Speaker 1:You have the sensor for the speed. So we have already the urban cities, urban areas monitored in Europe, let's say. But now we have already the urban cities, urban areas monitored in Europe, let's say, but now we have also the environment. Slowly you will have this monitoring, but it will be mainly based by satellite-based systems. So Earth observation, so it will be non-destructive, you know. So it will be non-destructive, you know, but we will have that monitoring and we'll have a full monitoring. We'll be able to really see if you know literally. I think now we already. I saw some I don't want to mention any names, but I saw some people told me now we, even with the public data, open data, not planet open data you can potentially track and detect even when the people cut one tree Everywhere in the world. If the people cut one tree, you could potentially, with scientific See that.
Speaker 1:You could detect it.
Speaker 2:Yes, which triggers a lot of questions on privacy. And what do you make of that when everything not everything, but let's say the eyes in the sky are monitoring, always, that's a bit of philosophical, ethical, let's say, discussion.
Speaker 1:For example, when you say today I personally, I really see it's very corrupt. You know, in general human culture we have this idea that we own the land and that we own the animals and like we are kind of like the gods right and we can, for example, imagine maybe 100 years ago, I don't know, somebody has a land and one day says, okay, cut all the trees. Right, and if you own the land, if you're some duke or something like, who can stop you? You?
Speaker 2:could do it. There's nothing.
Speaker 1:Even now in the world, you know, people can just go and cut forest and, uh, and they can say, okay, just strip all the land I don't know. Or or like they can say, okay, I want to build here, I don't know, some big pool, I want to put what I want to do some heavy industry, you know, and, and that was possible. But today in europe, you know you own the land. It's like you can check the regulations, you can check the law in netherlands if you own the, if you own some I don't know two hectares of land, and you have trees and you think you can just go and cut every tree. But you can't, you cannot, you need to have a, you need to have a permission yeah, and it's a, it's already yeah and also you need to make a plan also for if you have some land with trees, you have to.
Speaker 1:I think if you own it, you have to make up every five years, you need to management plan and things.
Speaker 2:Which is never enforced now, or very rarely. So you're saying, basically the regulation is already there.
Speaker 1:I think it was also this Never been easy to First. I find it more and more this idea that people own the land, that you own everything own the land that you own, and hold everything on the land and you are like God you can do.
Speaker 1:I mean it can go really bad and eventually you know like the ecosystem is the one. So everything you do on your land it has an effect on other people. So we are really into this together and and eventually I think in the future will be less and less, uh, it will less and less about even a discussion that somebody individually can decide about. You know, plants, animals, water, soil, whatever, uh, so so it it's a needs to be part of a watershed.
Speaker 2:Yes, it's kind of common.
Speaker 1:You know it's a. However, of course, you, if you own the land, you have the commercial rights you know to that land, but doesn't need to be the same, like ownership could be detached you can detach it.
Speaker 1:So it is. It is possible to combine that, but anyway, that's fascinating. I think there's going to be a total monitoring within 10 years. It will be for sure possible with technology. I was actually even thinking you know these financial benefits and the subsidies agricultural subsidies Before you know agricultural subsidies used to be a bit bureaucratic and you have to apply. You subsidies Before you know agricultural subsidies used to be like a bit bureaucratic and you have to apply. You wait, you know, and I think it can always speed up. And so, for example, whoever does in the future? Because if we talk about this European Green Deal, if you talk about energy and food production transition, then anybody who does it good, anybody who's champions, they should be really rewarded and they should be rewarded without bureaucracy. They should be very quickly.
Speaker 2:Which is not the case now. It's absolutely not the case.
Speaker 1:If you look at the cap subsidies, which is the biggest subsidy scheme on agriculture, even you have this paradox that you want these people, that they should apply and they should convince you that they are champions. But we can see. We have this Earth observation data, all these images, so we can see. I mean, you have, for example, you have also 30-centimeter and 10-centimeter images that exist from satellite, so you can see cars and even leaves. So you can see, you know cars and even leaves, you know, almost you could see.
Speaker 2:You can see the plowing. I think Maxar and this company.
Speaker 1:They specialize in the very high resolution VHR. So this is at the moment, I think it's 30 centimeters standard, but I heard that there's also 10 centimeters coming, and so what happens is that you know, you can see all this stuff and government should really locate these people that are champions. So anybody who does, for example, effective regenerative agriculture, regreening, building terraces, dams, land regeneration, restoration, anybody who does that, the government should come to them. We know what you're doing, we have tax benefits for you. We would like to put you on our website, I don't know, and all these things. They could really change the atmosphere. I mean people should really. The people are the champions of land restoration.
Speaker 1:I mean, what's the most important? When you think in the future, like from okay, what's the let's say, most important is we don't have to third world war and things like that, and we managed to get out of this cold war between these big nations. But let's say, beyond that, what's very important is you know ecosystem, climate, our soils. You know Soils are foundation of civilization. You know, if you cut a forest right I don't know if I said it already, but if you cut a forest, if you do a clear cut, you go somewhere in Brazil you do a clear, cut these trees. You know they need time to grow back. You know you could cut a forest immediately, plant a new forest and 60 years, you know you get 60, 80 years. You get the forest back, but the soil it takes 160 years for one centimeter.
Speaker 1:So if you cut the forest and it's in a place where soil can erode, uh, if you lose the soil and if I tell you, okay now, you take 10 000 years to get the soil back, you can forget it. It's game over. Who has 10 000 years to wait? You, you cannot go and oh, okay, I'll just go buy soil. I go to, I lost soil. Go buy how much soil, how much did you lose? I lost about 150 million tons of soil.
Speaker 2:You bankrupt, you bankrupt. Basically, I think that notion still is very little understood. And to go back to AI for Soil Health, we're halfway in and through like, where are we with the data cube? We're currently at the beginning of 2025. What is, let's say, a lay of the land or a photograph, if you had to take it now, like where, where's the work currently and and what is, uh, what is going on at the moment?
Speaker 1:yes, so so therefore, so health, therefore, so that you domain, you can check uh, I'll put. I'll put the link so, uh, we are, we are. We just passed 18 months, we had a review, 18 months review and, um, we and we're doing some really good progress. So some of the things we managed to produce data-wise we made a B-monthly time series of biophysical indices for Europe for the last 20 years.
Speaker 2:Okay, you have to explain what that means.
Speaker 1:So for example, we used the Lancet because we are interested in the last 25 years, so up to 30 years we're interested. So this is if you take the Lancet images you can derive some kind of a proxy. They're not like really measurements from the space, they are kind of proxy. For example, vegetation index NDVI. So that's a vegetation index. It estimates how much vegetation you have, how much leaf area you have, how much photosynthesis is happening. So you can estimate it based on the combination of near infrared and green band et cetera. And so you estimate that and this is one biophysical index. So one is, for example, the vegetation. Second is, for example, bare soil. If you know the vegetation, then if you inverse it, you know, you can check every month how much of the soil is bare. So when somebody does, you know, very intensive tillage, plowing, the soil will be bare. It's really no.
Speaker 1:You can see it and it will be zero vegetation, right? So there's no photosynthesis and you could say well, this soil was bare for like two months. And then you can do that for every pixel for last 30 years. And there is one index which is called tillage index. So how intensive was this tillage? And we now mapped it for whole Europe, tillage for the last 30 years. And I don't know. I think it's the first time there was some similar product, so the last 30 years, 25.
Speaker 2:And for how big are the parcels?
Speaker 1:So the pixel is 30 meter.
Speaker 2:Okay With.
Speaker 1:Copernicus, you could do the same at 10 meter. Yes, but Copernicus is only 2016, 2020, present time or 2017. And so you can do like a 30 meter and it's amazing, you can now look at the farms and now, if we have a pixel, what we do? We do trend analysis and we can say how much the tillage intensity changed.
Speaker 2:So you know where no tillage or no tillage has been used a lot even we don't know the farmers we don't know what they do, but we can see that somewhere.
Speaker 1:And we also put we added also Turkey and Ukraine because they're counted countries. So we have the whole European Economic Area, uk, we have the Western Balkans and we have Switzerland and Turkey and Ukraine, turkey and Ukraine. For all these places we have 25 years tillage, intensity, bare soil coverage, vegetation indices, etc. Then, using that data, we also combine it with this big monitoring programs of European Commission. One of them is called Lucas and there's a component called Lucas Soil. They did like a four systematic service, each survey about 20,000 locations in Europe. They take samples. We took the data and we run pan-European space time models to predict changes in soil carbon density, so that all soil carbon density, soil carbon stocks, and so we also analyze that to see how much people gain or loss carbon. So we also have that.
Speaker 1:This is a bit more complex because we need to fit models to estimate that. And there are. You know the soils are hidden right, like many things from satellite images, everything which is vegetation, buildings, water areas, floods. You just have an image and you can delineate, but soils are beneath, so they're a bit hidden and soils have layers right, so you have to also map them in 3D. So what we produce now, this soil health data cube for you, we produce 3D plus T, so 3D plus time, so you have space and you have the depth, you go to depth and you have in time. So it's a space-time data cube and it allows you to do all this trend analysis and this is all part of our general framework for soil health assessment and it's a very comprehensive framework. It is very new. I know other soil health assessment frameworks which are more like you come to the location and you take the samples and you send them to lab. You get back the results and say this is a soil health.
Speaker 1:Our framework is like I don't know, I don't want to be too arrogant, but I would say about between 500 to 1,000 times more complex, because our framework has seven steps and the first step is that we take these terabytes of data. Actually we just analyze 1.2 petabyte of Landsat images and we take petabyte of data and then we analyze these biophysical indices and we fit the trends for pixel. So we fit trillions of models to map these positive, negative trends and changes and to then first to see for any farm. We say first we look what happened here last 25 years. We want to know what happened.
Speaker 1:How much the Land cover change? Did you change your cropping systems? How much the tillage changed, vegetation? Do you have this primary productivity? Is it increasing or decreasing? Is it just noisy? You know it could be just noisy, just oscillations, but we see in some places we can really see there are negative trends and we can plot that trend of europe that you cannot see in this complex data. There's no way you can see. If you just look at this image somebody shows you you could see, okay, there's some changes. But when you take biophysical data, biophysical indices and you analyze the trends, then you can really do interpretation and you you can does it make you curious, like to go to, like some places absolutely, absolutely stick out.
Speaker 2:You're like, wow, I would love to go to this location in austria or this because we have now 25 years.
Speaker 1:If we have done this like 10 years ago, if you have on like, for example, 5 to 10 years of these satellite images, it's a bit more difficult to say whether something is a trend or you have, for example, drier, wetter years or something right. But when you have like 25 years and you see in this 25 years, maybe 20 bit up down, but in general it is going down let's say, the primary productivity, the, the, the gpp or the npp. If it's going down and you fit it and say this coefficient of negative change, it's statistically significant. So it's kind of like going to doctor. You know the doctor will say give me your blood samples, I analyze them and then after the test they say look, I look at your blood test, you have inflammation or you have some infection. You know probability, 99.99%, right.
Speaker 1:And this thing we can do now that's why it's called soil health, right. Also because we are doing some kind of we're kind of doctors for soil and for the land and what we are looking for, this 25 years of data. We're looking is there something negative happening, there's a negative process, and how much is it and where are the locations where it's negative? Likewise, you have also places.
Speaker 2:We have a positive process, positive so some people even more interesting if they're similar regions they can be neighbors they are next to each other.
Speaker 1:we had some places in Netherlands. We looked and we saw really they are neighbors, and one has a negative and one has a positive, so one is the champion and the other is, like you know, not really planning for the future, I guess. And so we see them next to each other and this is only possible because of this high resolution.
Speaker 2:uh, data, uh, and because you can go back 25 years, and because you can get, I mean, 25 years, which is not so high.
Speaker 1:Yes, yes, there's even better will be 50 years. Officially. There is data from 1985. There's a lancer data and what's missing?
Speaker 2:what? What do you need to analyze that? It's like cleaning up the pictures like, no like, because they had, they had the different systems.
Speaker 1:you know different sensors, so you need to analyze that. Is that cleaning up the pictures? No, because they had the different systems, different sensors, so you need to harmonize the sensors Because imagine there were a couple of research papers where they detected some changes, but the changes were not because of change in the nature, but they were changes because of changes in sensors.
Speaker 2:Changes in sensors.
Speaker 1:And that's what gets complicated. So as you go beyond 1997, 30 meters, it becomes more and more difficult. So you need to put a lot of effort.
Speaker 2:It's kind of a, and is that worth it or not?
Speaker 1:I would say give us a couple of years we might go back. You know we are like this movie. You know, back to the Future. We are, like many people, excited about the future, what's happening Actually now I look at lots of work we do. What I'm really interested in is to try to reconstruct, reconstruct as far as possible we can go to, really reconstruct objectively, not by people's stories, not by, you know….
Speaker 2:I thought this was it. So we have my parents, so really objective.
Speaker 1:So, based on satellite images, based on most objective data, aerial photographs, satellite images and we want to reconstruct, because if you can reconstruct last 50 years, if you can reconstruct the climate change, land degradation, pollution, you know, if you can reconstruct it, then if you fit the reliable models, then you can use this model to predict the future that's what's going to be my last question, actually, like has that future?
Speaker 2:I'm not saying predictions, but like, what do you tell, let's say, the champion farmer? You have the two farmers next to each other, the neighbors, and like how they don't even know.
Speaker 1:They don't know that some are champion, some, some people. They do things and they are very nice and they read a bit about the you know regenerative culture.
Speaker 2:They read about how to change some practice but how can we help them to do even more? Like are these models? Like not only looking back, which is fundamental, but you? You just went to a point of how do we then predict the future? How can we make decisions? We are very interested.
Speaker 1:We are a data science institute, spatial data science institute, but we are very interested in this social component. And so how do you? Because we make this data and we have now it's open data everybody can use it. Our mission is really just to promote people using environmental data and changing their culture really to change to restoration culture. That's really our mission. And so now we have this data, but we need to reach the people and we need to train them to use it and we need also governments and all these agencies to realize that. They have to realize there's these champions and you know we give awards to sports people. Somebody goes or wins the Olympics medal or something, or they play in the finals or something, and it's all in the news and you know the pictures and their girlfriends.
Speaker 2:That's our culture and their girlfriends and their TikTok channel's, our culture and their girlfriends and their tiktok channel, and I don't know and what about the champions of of land?
Speaker 1:what about the champions of uh land restoration? We should also find these people. We should uh celebrate them, uh, we should. We should say it's not, uh, it's scientifically, we have a scientific proof that they are champions, not because there's some political party or I don't know. But we say, no, this is, we are independently looking. And this is the top list. And you know, today you have in Europe, for example, netherlands, all the boundaries of all the farms. You can download them. It's public data. It's very nice in the Netherlands, so it's called digital cadastre, right, so you can download. I have it on my computer. It's like a two gigabyte file with the polygons and I can download it. Okay, I don't see the people that are owners. I don't know who the owners. But let's say, if I could somehow locate these people, I could send them all this data.
Speaker 2:You can select the champions. I could send them the data and say, look, you have really positive score.
Speaker 1:How do you do it? Tell us your story. And then you connect people, you know, to start interacting, to start exchanging. You know, and then we change this culture, that we celebrate these people right, that we celebrate the people who are most efficient in the green transition, that we really, both as a government, should celebrate them somewhere, officially, but also unofficially, through social media. You know, as long as people know, you know they will celebrate. It is a bit like sports, you know, and I am very jealous about sports because people are so much. Especially, I come originally from Croatia and sports is super important and I was always jealous. How are these sports people? Everybody knows about them. They know about their girlfriends, about habits, and they follow their social channels. You know.
Speaker 2:Why not land stewards and scientists? Yeah?
Speaker 1:but what about some good biologists or some ecologists that you know did something on the ground and you know there is our colleagues on the project from ETH. They made this Restored, echo Restored. Echo is kind of a crowdsourcing system for people to kind of register and say I'm going to do land restoration on my land, and then they provide help. You know, they provide some tutorials, I don't know and they also, they want to similar like what we do.
Speaker 1:They want to track, you know, and see whether they want to. Similar like what we do. They want to uh track, you know, and see whether they really do that. Uh, land restoration, yeah, um, and they, they are apparently. I think they have now about 200 000.
Speaker 2:I don't take me officially, I'm just yeah, yeah, but there's something, it's really growing and they were very excited.
Speaker 1:We were in vienna now we just had this global workshop on uh, on a open net monitor project, and they were there, they were presenting and they say, yeah, you know, we people are interested, they would like to sign up and uh, you know, it could be like a facebook for land, land book or something, I don't know um and yeah, but it's an interesting culture shift.
Speaker 2:Um, like, how do we celebrate the people that are, like we have the top 50 farmers now, like, how do we make sure, like we celebrate chefs almost more than than farmers, which is which is crazy, yes, and how do we? How do we? And, of course, the sporting side, etc. And people like you as well, you're in the top 0.01 percent of um of scientists and it's just not recognized in our world 0.1 percent.
Speaker 1:Most cited. I'm just the most cited doesn't mean that I'm a high quality, you know. Just people cite me there.
Speaker 2:There is a connection there. Let's say, let's leave it to other people, but you're let's say you're well established in your space and you know I'm not talking about so this is just the land, right, but listen in the but in the green transition, but also people that, like, for example, uh, switch to the, uh, move away from gas heating, uh, solar, solar energy, solar panels you can see that exactly.
Speaker 1:Recycling of materials, all of this stuff, you know we should, uh, we should have some way to track that and to celebrate these people especially that are very good at it.
Speaker 2:It's interesting. You make the point of celebrate, not punish, the ones exactly exactly. It's very nice, especially that are very good at it.
Speaker 1:It's interesting you make the point of celebrate, not punish. Exactly, I think it's a very nice. It's a very good cultural difference.
Speaker 1:You can also, I think when, I told you this full monitoring system, let's say, within 10 years you have a full monitoring. It might happen. I just spoke with somebody from European Commission and I said what, what about? You know, uh, somebody makes a spiel or something. Um, you know some garbage spiel, or. Or somebody does something which is not in the plan and you can see it on satellite images, like when you, when you drive a car, at least in netherlands in germany be different, but in netherlands they just send your letter, pay right you've been driving with his feet.
Speaker 1:Yeah, I drive sometimes like uh, they are very precise. In netherlands I drive like seven kilometers above, I drove like 107 and they allowed 100 right and and I get I think it's 35 euros and 40 euros bill and there's like this is the bill and that's it, just pay right. And so it could happen also with environment. Environment, you know you could, so you're not driving the car, but you know you have a land and you didn't, you're driving the tractor you, you promise you're going to do, let's say no tillage or something.
Speaker 1:But now suddenly we see you doing tillage and you get the fine.
Speaker 2:You know and that I think we get into. I mean, that's a whole different conversation which is for another time, but it gets into like how, I think when we reward people for like it's the difference on practices versus outcome, and like if people, because even tillage to a certain I mean, there are ways of tillage that sometimes are beneficial, there are ways of soil disturbance that are not completely horrible. There are, like it's always a bit more nuanced. Of course we don't want like a tillage fee, that by by by, but if, if it harms the soil health, if it harms the production, productive quality, etc. Um, it might need to, but I think I I love the celebration piece, but we'll need both, because we have a lot of these rules. We have a lot of these and it's never, or almost never, reinforced.
Speaker 2:On methane, on spillage, on pollution on rivers, none of that, or very little of that, is actually being monitored or let alone even being fined. So there's a whole world there to discover. But I want to be conscious of your time as well and wrap up this check-in one in 18 months after the start of AI for Soil Health and almost three years after we checked before, and of course we'll be checking in with you as this project comes further and as well, I think, the world of AI like. Even if we talk, six months from now, a lot of new things will be happening, even though you said I'm excited but not so excited about the hype.
Speaker 1:I still think it's a hype. There's still. We have to see there's a lot of noise also being generated and it creates an illusion.
Speaker 1:Even it damages some people, creates an illusion they don't have to do the work anymore, and also it can be misused. You know there's more perils of AI technology, like especially the deepfake videos to manipulate people, especially by governments and things. Ai technology, like especially the deepfake videos, and to manipulate people, especially by governments and things. I do feel like there's one thing I didn't have time to talk and I would like to mention it. So you have this satellite technology. You have also the drones. Today you can buy a drone.
Speaker 1:You know Probably there will be soon. Like you know, tractors will go automatically. It will be like a robot taxi. You know, uh, probably there will be soon. Like, uh, um, you know, tractors will go automatically. There will be like robot, like a robot tax. You know so lots of things that they're, they're coming. But there's one thing in air for soil health we discovered, uh, it's a kind of a bottleneck. We need more, more like a way, cheaper lab data. So this is like somebody is on the field and they take a sample of the soil and this is something I was thinking is going to take off, but it's actually really stuck and, like I was thinking something like analysis, you take a little sample of soil.
Speaker 2:Yeah, we've had some companies like that you pay 100 euros.
Speaker 1:you know you pay 100 euros and you get everything. Basically and I was looking at the prices no way it's very expensive still, it's also bulky and this whole spectroscopy also had a big hope because the same stuff you have from space, you have the multispectral data data and now you could also have this, uh, even handheld sensors, you know, to scan uh the samples on the ground, uh, but also that is is highly complex, it's still not uh moving.
Speaker 2:Yeah, we had, we had, I'm gonna put it in the show notes. We had somebody on with a drill based system, um, in the us, mainly very promising, but absolutely not there yet. And and there's some others yes, I don't. Also don't want to mention names.
Speaker 1:I know most of them. You know I don't want to mention names they do say always so we have a solution, this is it, uh. And then you look at the cost and the costs are way, way up. You know it's, it's not something you will do like, uh, massively, you know, and, and just somebody who has a small chunk of land, you know small, small, whole farmers, uh, you know they would not not get into it, so that's. But you know, for example, it's a very it's really pity because there's a is a person. If you want to do the blood test, you know you'll, you'll pay some small amount and next day you get the results. Right, that's, that's. So when do we? We want to have a blood test for soil that you get the next day results, and it could and it can feed your model and it costs like 100, like 100 bucks or something and that's, that's still.
Speaker 1:Uh, that's a way, way and I I'm thinking who is responsible for that, why we're so late? To that. It is a bit shared responsibility entrepreneurs, because I think they have two focus.
Speaker 2:Labs as well. We've seen many labs. You send the same sample to different labs. You get different results.
Speaker 1:If it wasn't for European Commission I'll tell you now honestly, if it wasn't for European Commission, they did this Lucas soil, this project, pan-european soil projects would have been so difficult because if it's so expensive, right, and you have nothing to start with. It would have been this paradoxical situation that we basically, you know, have all this data, data, satellite images, serial photos, and we have no idea about the soil. We actually, if somebody asks like, do you have a microplastic soap? No idea. Do you have some pollutants? No idea. What's your solar organic carbon? No idea. So we would have been really tabula rasa. Uh, if it wasn't for this lucas soil, and even with this lucas so it's still limited, I mean, it's still, it's a peanut. You know how much big is the value of the land and the agriculture, products and things. It's really like it's per mil, so one millionth of the cost, you know, and so that's very.
Speaker 1:It worries me because we have very little ground, data in situ and also there's no solution. I actually think this Air for Soil Health project it might even finish. It will finish in three years. It might finish and we still won't have something that we give to the farmers to say, hey, you want to do your soul blood analysis? You know, uh here and you can do it, it's inexpensive. You should absolutely do it, you know. Now I cannot tell them that if I say, they will say you told me and I have to spend, like I don't know, 50 000 euros and uh, and that's not, it's not going to work that's not gonna happen.
Speaker 2:No, let's. Let's check in and see where we got. I think there's there's work being done there. That's not going to happen. Let's check in and see where we got. I think there's work being done there. But it's very tricky to do that in-situ, in-field.
Speaker 1:There is technology, but somebody has to put that technology, to make it operational and to make it easy, make it cost-effective.
Speaker 2:Make it cost-effective massive usage, and this thing I don't see it's not happening thank you so much, tom, for coming on here and sharing good luck with your podcast.
Speaker 1:We always love to watch in the weekends and we did so much interesting stuff on your podcast so many interesting characters, very personal stories. We also share it with our friends, by the way, who are not specialists, but well done. So also kudos to you, kun.
Speaker 2:Thank you so much. The European Union's Foundation for Environmental and Environmental Health has funded the European Union's funding for the European Union's research and investment in the environment. Ai for Soil Health is funded by the European Union. Thank you so much for listening all the way to the end. For the show notes and links we discussed in this episode, check out our website investinginregenerativeagriculturecom. Forward slash posts. If you liked this episode, why not share it with a friend or give us a rating on Apple Podcasts? That really helps. Thanks again and see you next time.
