Drones on the Farm: How Can We Improve ROI?

Exploring how farmers can and will see a return on their drone investment is a topic I’ve discussed with farmers as well as experts who are working to develop new solutions that can be utilized on the farm. Being able to get specific around exactly what drones can do for growers is increasingly necessary as the talk around the potential impact of the technology continues to create headlines.

Laying out that sort of detail is something Dr. Antonio "Ray" Asebedo is focused on doing for people on every side of the precision agriculture space. As an assistant professor of precision agriculture at Kansas State University, he’s heavily involved in his departments’ efforts to work with farmers and stakeholders directly.

After getting his undergraduate degree in agronomy, he went straight to a PhD in soil fertility that was centered on developing nitrogen recommendation algorithms for optical sensors, predominantly focused in cereal crops. He worked with a variety of remote sensing platforms while also working directly with farmers to help create algorithms that provide them with actionable data, which enables them to make better decisions.

Dr. Asebedo talked through how thermal imagery is creating possibilities in nutrient management during his presentation AirWorks, and I was able to catch up with him to discuss some of the insights he laid out there along with the challenges he’s encountered around drone adoption on the farm, plus plenty more.

 

Jeremiah Karpowicz: What’s the most significant aspect of the precision agriculture program at Kansas State?

Ray Asebedo: The fact that we're working directly hand-in-hand with farmers and crop consultants has really helped the development of algorithms within our program. It’s kept us grounded on what is actually needed and necessary.

These are people who have an immense amount of expertise and experience themselves. They've been at it for decades. They know their soils and they know their land very well. My job is to try and help quantify that knowledge and implement it into these algorithms. We want to teach these machines how to be like human agronomists.

 

Generally speaking, is that something farmers have or are willing to embrace?

Well, that brings us to the cultural aspect. Farmers know how to farm their ground, and they've been successful at it for a long time, and this idea that we don't need their info or their input puts them off a little bit. We need to recognize the value and expertise of info that they have.

Unfortunately, that’s a perspective developers don’t always take into consideration.

When we're designing these management tools to help farmers, we need to consider what information they need and how they’re willing to utilize that tool. If we just consider these technologies from a developer perspective, then often they don't get adopted.

 

What kind of challenges with adoption have you come across?

Let’s look at yield monitors as a good example of this issue. Yield is the sum of all the interactions between farmer management, genetics, and the environment. Therefore the concept of yield monitors and their potential for improving farm management makes sense. Yield monitors have been around for along time now, but how much are they actually being utilized to improve farm management? The truth is not very much.

The roadblock is that just because you can get some yield data doesn't mean that you can innately understand what's happening behind the scenes with the environment interaction. There's also not a streamlined algorithm to help them process out that yield monitor data to tell them what it actually means and how they should consider adjusting their management plan based on what the yield monitor data is telling them.

 

Are those roadblocks related to the information that tools like yield monitors are providing, or around how they’re providing it?

We need to give farmers recommendation options to consider rather than data they need to sort out.

A big problem I've seen with tools is around having a lot of steps and not having a streamlined process to take the farm or crop consultant to a direct answer. They're not going to waste the time to try and utilize a tool that just provides data, because they've already got too many things going on at the farm. They don't have the time to learn how to utilize these tools. If the tools are not intuitive to give them direct answers so they can make a direct management decision right then and there, then whatever the tool is has a tendency to fall to the wayside.

Preprocessing data for the farmer so it gives them specific options or answers is the key. That's really what they're looking for and what they're asking to get. They want to know what they should do next. Yield monitors are just providing data, and that’s the reason they haven’t really been utilized to improve farm management. Farmers ultimately want the recommendation based on the data, not necessarily the data itself.

 

What does that tell you about how drone technology can and should be positioned in the precision agriculture space?

We need to learn from history. We need to look at what technologies have come down the pipeline and study their success rate around adoption. If it was a poor adoption rate, despite it having a lot of potential, why did it not get adopted?

Specific to drones, the first thing farmers ask me is, "what can I actually do with this?" They want to know if it will actually give them their nitrogen recommendation map, and then how easy is it to get it. If I tell them they have to create a mosaic and they have 15 other steps to do to convert it from imagery to a nitrogen recommendation, they just say to forget it and walk away. They’ve got too many other things to do.

We need to develop drone technology to be within the design parameters of what a farmer is willing to do. It comes down to the same old thing: is the system easy to use and does it give the farmers a direct answer they need?

 

I can certainly understand the desire and appeal around being able to simply push a button, but is that something we’ll be able to see for the precision agriculture industry as a whole?

Ultimately, what I see is that you're going to have algorithms that are crop specific and environment specific.

Take my home state of Kansas. I could have an algorithm to give nitrogen recommendations to wheat farmers in southeast Kansas, but it would be completely different from the algorithm that's being implemented in southwest Kansas. Although they're both growing wheat, it's a matter of environmental differences. If we're trying to develop a system that maximizes profit per acre and reduces environmental impact, we have to make them site specific. It has to consider the different crop physiologies. It has to consider soil by weather interactions that take place.

That’s also where machine learning comes in. We could have the same base algorithm with wheat and corn and different crops, but over time as it's being implemented in southeast Kansas or Nebraska, it starts to learn and evolve so that it can optimize itself for the environment it's working in.

In the big picture, I could see that we might have the framework for an algorithm in an AI where it could be the same across the world, but how the internal components work is likely to be different, because it will have adapted to the environment it's working in.

That's where you start thinking about things like what you use your yield monitor data for. The algorithm is going to start using that yield monitor data, so then it can learn and adjust itself to optimize for that specific field. When we reach this point with these technologies, pushing the initiate button may be all that’s needed.

 

Your presentation at the AirWorks event was focused on the possibilities for nutrient management that thermal imagery is creating in agriculture. Briefly, what can you tell us about those opportunities?

Drones are very conducive for crop monitoring. It's easy to put a drone up in the air and fly a field. Because of that, whether we're talking about thermal imagery, multi-spec or regular RGB, we start getting more data over time at a tighter time interval. That allows us to start seeing the onset of any sort of nutrient deficiencies sooner. Because of that, our detection and response time to crop stress is improved. The window for detecting and responded to crop stress is typically narrow. However, with this technology, we'll be able to address the problem sooner which will make it so that we might not lose any yield at all. We typically lose yield if our response time is too slow.

Not only that, but we can also see all the variability within the field. We may have sections of the field that have nitrogen deficiencies and then another area of the field that has more than enough nitrogen there to carry it for the entire season. If we can quantify that, we can create variable rate nitrogen recommendation maps that match the site-specific needs of any given crop field. In addition to given the farmer variable rate nitrogen recommendations, drone can tell them when is the right time to apply the nitrogen. The drone can inform the farmer of the right time and rate to apply nitrogen to maintain nitrogen sufficiency throughout the growing season.

The nice thing with the variable rate nitrogen recommendations is that we're only applying nitrogen where it's needed. So other areas of the field that have been given nitrogen freely by Mother Nature can be left alone. You can be sure you’re getting an optimized nitrogen recommendation that's going to give a high yield and a high profit per acre while minimizing environmental impact.

Thermal imagery is unique, it can be used as a proxy for crop root conditions and its ability uptake nitrogen. This also ties into how thermal can impact water management. Nitrogen moves by mass flow, aka water, so you should consider the two a packaged set. As your algorithm is assessing crop nitrogen status, it should be using thermal to assess water status and heat stress. Therefore, you should be using thermal for water management also. Running a field scenario where thermal imagery is detecting heat stress, if you know there's plenty of water in the soil profile, your algorithm can list out the factors around why heat stress is occurring despite ample water availability. At the same time, we could be seeing areas of the field that have sandier soil or less water in the profile, and we genuinely need to make a water application if irrigation is available. Then we can start doing variable rate irrigation maps and start conserving our water by only applying it to crops that need it. It's the same concept that works with the variable rate nitrogen, just applied to water.

 

 

That kind of extrapolation ties into other opportunities that drone technology can open up on the farm, doesn’t it? I understand that’s something you’ve personally begun to explore with a tool for the cattle industry.

One thing to think about in terms of drones being used on the farm is around how to improve return on investment. How can we improve ROI? The more things that we get the drone to do on the farm, the higher ROI it has across the farm.

Most farmers in the Midwest run diverse operations. They have a crop side and a livestock side. When I was out there working with farmers on these nitrogen recommendation algorithms, they saw the success it was having and they were impressed. Then they always asked about what I could do for their cattle. It just got to be such a recurring thing that I realized I should do some work with drones and cattle. I based my development on them saying what they actually needed. We have some interesting drone cattle applications in the works and I look forward to showcasing them in 2017.

Developers really need to start thinking horizontally across the farm operation. What other areas of the farm should they be targeting for development and improving profits?

 

We talked a lot about the barriers that exist for farmers and developers around seeing drone technology actually adopted, so what are some of the things people on both sides of that conversation can do to break down those barriers?

The best thing to do is communicate. Developers really need to open up lines of communication with the farmers. They need to be working with farmers directly. The thing is, being able to do that means you have to deal with the cultural barrier.

Farmers typically want to meet you in person. They don't want to just get a phone call or email. They usually don't like that because they want to know who you are and what you're about. Basically, they want to gain that trust. It's incredibly important in the ag community.

On the farmer side, they need to be open-minded and realize there are a number of individuals out there who are genuinely trying to develop management tools to truly improve their operation. They're not just trying to make a quick buck or push something on them that doesn’t work.

That kind of communication is essential, and the thing developers need to think about is how often are they actually making that effort? Have they gone out to the farm and dealt with the farmers directly on a repeated basis? To make the farmer feel that that their input is actually valuable? Developers that do that almost always come away from it thinking how it was very worth their time.

After that trust has been established, word will spread in the ag community, and these are people that will continue to work with you the rest of their lives. That's when you’re seeing a real relationship because they're helping you and you're helping them.

If you want to develop things for the farm, you've got to step foot on the farm.