The future of farming

0:00This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. Since we started this show eight years ago, it's become an archive of amazing and impactful

0:08David Lobell:work by my Stanford colleagues. Research is not something that just happens in the lab, and as you'll hear on this show, the research at Stanford can impact areas like health, technology, law, and business, and many other topics that can affect everyday life. We hope you'll tune in to learn more about how research has the potential to help your life and to help the lives of people you care about in your family and your community. It's true that our society in general is much more well-fed and better off. The last 25 years has not been a success story. We've seen reversals in a lot of the progress.

0:42David Lobell:We see the real price of food has gone up. We see the rates of deforestation have ticked up. These scenarios, again, for climate stabilization are requiring a super optimistic scenario for agriculture, and we've headed in the other direction.

1:00This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. If you're enjoying the show, if you like what you hear, go ahead and follow it. It'll guarantee that you never miss an episode. Today, David Lobel will tell us that the production of food is not just about getting stuff for us to eat. It also is intimately related to political stability, international war, and environmental damage. It's the future of farming. Today, we're going to continue our feature that we call The Future in a Minute. I'll ask David at the end of our conversation some quick questions. He'll give us some quick answers.

1:34And that'll be the future in a minute. Also, don't forget to follow the show so that you never miss the future of anything.

1:47So we all know a little bit about farming. Of course, some people know a lot because they're farmers. But for the rest of us, we know that you plant stuff, you water it, you fertilize it, it grows, you harvest it, you send it to market, and people eat it. But food and food production is a lot more complicated and is actually related to things like political stability, international conflict, and environmental damage. So it's a big deal. We also know that farming comes in many shapes and sizes. There are small family farms, and there are huge, large industrial farms. So what is the state of farming?

2:23How are we doing? Are we feeding the world? Are we doing our best? or is there a significant room for improvement? Well, David Lobel is a professor of earth system science at Stanford University and an expert at agriculture and food security. He uses satellite imagery, data, and computational modeling to understand what's going on in the world, over 500 million farms, in terms of productivity, the crops they're growing, and how things are doing. They use that data to come up with new hypotheses about how they can help farmers be more productive and ultimately more happy and get food into everybody's belly.

3:01David, to start out, why have you devoted your research career to understanding agriculture and food security?

3:08David Lobell:I guess it started, I was a student looking for something that I thought was interesting and important. And I was a math major. So I was looking for quantitative stuff, a lot of interesting stuff, but I didn't view it as very important. And similarly, some important things, but I wasn't really fascinated by it. And I was very interested in environment. So I went to grad school in the environment and I slowly sort of came to the conclusion that the most important thing in the environment was how we were going to feed the world going forward. And that got me into the field and that got me more fascinated.

3:38David Lobell:And over time, I just sort of understood deeper and deeper why this is such both an important and interesting topic. As you look back on your childhood and raising, does it kind of make sense given the kinds of things you did or the kinds of experiences? is do you have a farming or agricultural background? It's just a very interesting decision. Yeah. No, I grew up in the suburbs of New York City, so it's about as non-agriculture as you can get. You know, looking back on it, I think I can see sort of indications of sort of interest in sort of – or being taught sort of that it's important to help improve the world, try to think of who the most impoverished are and how you can help them.

4:18David Lobell:That was sort of inherent in the sort of culture in my upbringing. But it wasn't really till I, like I said, I got into the field and really understood agriculture that I really sort of got turned on to it. Great. So one thing I wanted to ask for kind of as a ground landscape setting is farming. We all kind of know what farming is, but I think we could all benefit from a little reminder of how you think about farming. We all think about big industrial farmers, but we also know that there are people who are family farmers. And as you look at this in your work, which one do you focus on and how do you maybe combine them?

4:57Because they seem to be very different.

5:00David Lobell:Yeah, our work is really global. So we work in some very developed countries, some very developing countries. The specifics of agriculture always change where you are. I think there's kind of two general questions that people have. One is just how can farmers be more productive for their own benefit? So how can their yields improve? How can they reduce costs? And you have that question throughout the world. The other set of questions is about how can you improve sort of the public impacts of farming, basically maintaining farmer livelihoods, but nudging them towards things that are better for the environment and better for society.

5:36David Lobell:And so when we're looking at problems, we're typically working with either people who are setting policies or making investments or people who are talking directly to farmers and trying to give them advice or provide them with services. we're very rarely working with the farmers directly but we do talk to them a lot to get an understanding of what the sort of real problems are so small holders large holders um they all have sort of as i said their peculiarities but they're all fundamentally dealing with the fact that it's a very difficult way to make a living it's a very difficult uh system to understand how can you change and be a little bit better great so now um in terms of farming what would you say You gave us a little bit of hint of this, I think, in your previous answer.

6:20But what are the big issues in your world that are pressing and need to be solved?

6:28David Lobell:So if we take this dichotomy, if you look at the sort of more developing countries, especially in Africa, the big issue is just how do you get productivity higher? We know that most of these areas are 50 % or less of what their crops could be achieving with good management. but there's all sorts of good reasons that they don't achieve that. And so there the real questions are, for example, what kind of fertilizers do they need? Whether planting trees would be helpful. Are there soil amendments that could help? There's been, we've done work in India, for example, on moving from just hand broadcasting of fertilizers to spreading it with a handheld like rotary thing.

7:07David Lobell:So it's the specifics vary a lot, but it's really about, there's always hundreds of ideas that people have about what should work. There's usually a very small fraction of that that actually will work. And we want to know how well things work and where they work. On the more developed side, it's really, I think, increasingly about getting agriculture to a place where it is, some people will call it sustainable or regenerative or net zero, but trying to basically really reduce the cost that agriculture has to society. So in Europe, for example, there's lots of policies being promoted to increase the biodiversity of agriculture, try to reduce the nitrogen going into it, try to reduce the carbon emissions from agriculture, or even try to sequester more carbon into agriculture.

7:52David Lobell:So this is sort of under the umbrella of, as I said, regenerative or sustainable practices. The U.S. also has been promoting a lot of these. And the key is you want to promote things that help farmers as well, or at least don't hurt them. Otherwise, they're not going to take. And so there's a lot of, again, a lot of ideas about what should work, a lot of things that on paper look good. And then when they run into reality, they don't work as well or they work differently than we expected. So I'm really interested in your statement that the African farmers may be at only 50 % of what they could do.

8:23And first of all, how do you even know that?

8:27David Lobell:Well, you know the seeds that they're planting. So you know the genetics of the seeds. you can run experiments like controlled experiments in those regions where you give it everything that you know you would think it needs and you can see the yields are much higher you can also use sort of basic understanding of biology and biophysical models that that simulate and you can cross-reference all of those and you can also see that there are some farmers actually achieving two or three times the yields of their neighbors and so you can get a very clear sense that things are not going well you can also just go into the field and see all the problems that they're having And can you tell is this – so I'm sure that they are – that they would like to increase – I'm suspecting they would like to increase their yields unless there's like perverse incentives that you sometimes hear about.

9:11Is it a problem of resources that they have to actually enact these improvements or is it lack of kind of dissemination of these good ideas or are they structural barriers to actually taking them up for kind of quirky

9:26David Lobell:local reasons? It depends. And you can imagine there's vast literatures and all those things arguing for each of those. If you were somebody who really believed in institutions as the barrier, you could say, we have all the technologies we need and we just need to get them out there, or we just need to build better infrastructure, better information systems. But the reality is, you know, farming is extremely hard. And when you're in a, for example, a tropical soil that is very acidic, has very low organic matter, a lot of the things that work on the experiment stations or that work in other countries, don't actually work.

9:58So it's a very empirical question about what actually works and what works

10:02David Lobell:at a good cost level that you can actually justify it. The other thing that's going on in those systems is the land holdings are so small that even if they knew something with enough effort and enough inputs would be dramatically better, the returns on that investment just don't really get them out of poverty. So they're better off spending their time trying to get get work in the city or, or, you know, find a better job. So there's all sorts of complicated things, but, you know, given the world we're in, you know, we're trying to look for the things that on the margins can help kind of get the system going or, or improve things.

10:40David Lobell:And there are a lot of creative ideas out there. There are a lot of NGOs and governments trying things out. And what we're trying to avoid is them taking 10 or 15 years to try things out. Because again, the traditional kind of approach is to learn very slowly. And, you know, a lot of our work is about speeding that cycle up and really in almost real time understanding, okay, this thing really looks promising. These other five things don't. Now, I do want to get into some of the, forgive me for this phrase, but the cool technology that you guys use to gather data, to try to learn what's happening on the ground, and then even propose new initiatives.

11:18So can you tell me how your kind of lab works in terms of gathering the kind of raw data that allows you then to make inferences about the state of a certain region, either it's poverty or it's crop production and the efficiency of all of that? Sure.

11:35David Lobell:I think the starting point is that the traditional approach, like in a lot of sciences, in agricultural sciences, you design a hypothesis, an experiment to test that hypothesis, and then you run the experiment. And that is how agriculture has proceeded. Again, the challenge is there's so many factors involved. There's a lot of heterogeneity, like different effects in different places, that even if you take the decade to run a whole bunch of experiments, you only have, you know, some ability to draw conclusions. So our basic approach is to, you know, observe the experiments, like understand what they're saying, but take a different approach, which is more of a big data approach of just observing what's happening.

12:11David Lobell:And the key here is that agriculture, whether it's small holder or large holder, is an outdoor activity, right? We can actually use satellites to look down and see what's going on. We can see how the crops are doing. We can often see how the humans are managing the crops. Not that we can see the individual humans or their tractors, but we can get a sense of whether they've tilled the soil, for example, or whether they've planted this crop or that crop and what kind of sequence of activities they did, whether they've irrigated, whether they've planted trees, all sorts of things. So we're just kind of observing.

12:44David Lobell:The key here is that there's 500 million farms around the world. So we can go from a space where we have hundreds of very carefully controlled experiments to tens of millions of more natural experiments where we're just observing and then we're seeing as things come in and out what were the effects and so our lab a lot of it is focused on how do we observe the things that we want to observe how do we make sure that we're actually measuring things well enough and so a lot of our field work for example is in taking ground measures to make sure that the things we are detecting from satellite or other sensors is good or good enough to do the analysis.

13:21David Lobell:And then the rest of our lab is really not so much developing causal inference techniques, but trying to use the latest and greatest causal inference, meaning, you know, using observational data to make causal statements. And doing that in a context where we know all the data sets we have are not perfectly measured. So we have to be careful about sort of interpretation in that context. So I would say the cool aspect of what we're doing in terms of the techie innovative stuff is that we use a lot of cool sensors that are at this point kind of on a daily basis observing all the farms in the world.

13:55David Lobell:And so we can build massive, massive data sets that we can then mine for insights into what is working or what might work. It's so attractive, as you know better than me, because we think of satellites as looking out into the world, into the universe and collecting stuff about the stars. And in a funny way, you've just kind of turned them around and said, hey, let's see what's going on on Earth. So it is very attractive. So let me just ask a couple of kind of basic questions. Are these satellites up there for you and your colleagues and farming and food people? Or are you piggybacking on other satellites that are there for other reasons?

14:32David Lobell:Well, actually, some of the early motivations for satellites back in the 70s was that we were trying to figure out how much grain Russia was growing. because there was a couple of episodes where the world markets were really surprised that there was a shortfall. And that kind of motivated some of the initial ideas of what's called Landsat, which is the first major satellite. But if you fast forward over time, the satellites became used to study everything. And many of the new sensors out in space are designed to look at things like carbon cycling, forest deforestation, lots of hydrology questions.

15:05David Lobell:So we often find ourselves now looking at sensors. There's a sensor, for example, that measures gravitational changes in the Earth's surface. And there's other sensors, for example, that measure with lasers. They measure the heights of forests. But we find ourselves often repurposing that to look at agricultural questions that they weren't designed for. Basically, we have lots of sensors measuring different things at different frequencies, different resolutions, different temporal kind of cadences. And that those all have something to say if you are asking the right questions about the agricultural system.

15:40Great. And just a couple more, because, of course, I'm intrigued by this. What could you give me like a litany of the types of things you measure and also like how precise both in space and time? Like, are we get are we I think you kind of implied that you're almost getting a daily update on each farm. But I don't know if you meant to say that or if I heard that wrong. And then also, what kinds of things would you look at for a given farm? You made some reference just now to some of the things you can sense. But I'd love to just get an overview of these kinds of data sets.

16:13David Lobell:Yeah, I also have to be careful in Silicon Valley, not confusing the aspiration of where we want to get to where the reality is right now. I think the reality is that we can see large fields every week in most parts of the world. but in the last five years we have gotten to the point where the both the temporal frequency is getting down to the daily time scale and the spatial resolutions of these are getting down to very very small fields but fundamentally all you're measuring in these cases is light reflecting off of the surface or you're measuring like a lidar or radar beam that bounced off the surface so what we have to do then is translate that to things we care about and there we've made a lot of progress on something.

16:55David Lobell:So just to give you an example, I was, you know, this week working with some European data. So the European consortium, the research consortium now not only produces great raw satellite data, but they're producing for every field in Europe, what the crop was grown this year, whether or not they planted a secondary, like a cover crop, what time of year the plant was sown, when it was harvested. We've done a lot of work in our lab on measuring yields, so productivity at the end of the season or even a little bit before the end of the season. And we've done this for a bunch of different crops. So I would say that those variables, the ones I'm describing, we have nailed pretty well.

17:36David Lobell:And we continue to test them in the really complicated smallholder systems where it's still a bit of a challenge, but they work as well or better than a lot of the traditional ground-based measures. And then you get kind of to the more complicated, like the nuances of exactly how are they preparing the soil exactly you know when are they are they growing more than one crop in the field what kind of trees are on their landscape and and you know what kind of effects those might have so there's a whole suite of things where when students come and they're looking for like real the technical frontier that those are the things that we work on and a lot of the challenge there is actually getting good quality ground data at scale to test And so we partner often with research partners, either in private sector or public sector, to get that grounded.

18:19David Lobell:We don't really have the capacity here to do it all ourselves. Are you able to tell what exact crop is growing? Like, is the colors or the texture enough to say that's corn, that's wheat, etc.? With a snapshot, it's hard. So there are some things, like if you have a rapeseed or canola field that's in flower, like you can see it's very yellowing. But that's typically not what we're using. We're typically using the fact that crops have slightly different seasonalities. They also have slightly different levels of peak biomass. They have different leaf angles. So for example, soybean with very flat leaves is going to look quite different than corn because of the way it reflects light.

19:00David Lobell:And then there's some interesting things like sunflower, actually the way it tracks the sun during the day, there are morning and afternoon sensors. And so you actually get a good contrast for the sunflower plant of how it looks in the morning and afternoon. So there's all sorts of tricks that people have used. But fundamentally, if you have a field growing one crop, we're pretty good at telling what it is. If you have fields growing multiple crops, that's more difficult. Yep. And going over a lot of the papers that you've published recently, I saw that you're building what they call an AI, these foundation models.

19:30And I'm guessing it's like models that can be used to build all kinds of apps to do these kinds of analyses of temporal and spatial distribution of plants. This is The Future of Everything with Russ Altman. We'll have more with David Lobel next.