Joshua McGrath
Extension Specialist,
Agricultural Soil Management, University of Kentucky
Joe Luck
Associate Professor of Biological Systems Engineering,
University of Nebraska-Lincoln
Pictured Above: Despite different objectives within the academic realm of precision ag, Josh McGrath (left) and Joe Luck (right) were able to find some common ground during the 2018 AETC –– dissecting the void in delivering reliable, repeatable on-farm data through trial and error.
Josh McGrath: When you talk about the needs of farmers with precision ag, from my perspective on the soil fertility side, it’s a question of, ‘At what resolution do we make our recommendations?’ and reducing that knowledge gap. Knowing what we don’t know is a big problem with precision ag at the end user level because they are being sold on capability. We have this ability to vary seed, vary fertilizer or vary crop protection. The engineers have done this great job of making this happen. And then someone has to sell that product. Sometimes we extend beyond our knowledge on the agronomic side of how to vary any of those inputs, whether it’s seed, fertilizer or crop protection. Those are the big things.
Joe Luck: The biggest problem I see is a lack of educational and baseline research. We’ve talked about that before. Is everybody doing their field studies the same way? Are we varying things to cause and effect? Are we analyzing the data so that we actually know that we have a statistical response. That’s a little bit beyond my experience, but to me, those are some of the things that, if you don’t get it right, you’re not making the right recommendations and the equipment can control things down to a certain very, very small resolution. If we’re not going to keep doing this applied research, and do it so everybody is comparing apples to apples, how do we bridge that gap?
“In its current state, yield monitoring is not precision ag. It’s good enough to give me big huge coarse trends in yield across big distances…”
– Joshua McGrath
McGrath: So going down this rabbit hole of interpreting on-farm research data is an interesting question. We need some other people that have expertise in that area. You’ve got a soil scientist — who for his grad work had to do some statistics — and an engineer who had to do some statistics, but we’re not statisticians and there’s not many statisticians working in this field. One of the things we need is expertise.
If we’re doing research, we’re trying to save money. I can’t afford a Ph.D. in mathematics. They command high dollars and the farmers certainly can’t do it. At best, the farmers come up with an average, they might do four reps and they’re just looking at the average. It’s that standard deviation that matters, that measurable precision. Our confidence that we’ve converged on the mean is what that confidence interval tells us.
Luck: And a big challenge is the size of the equipment.
McGrath: Absolutely. If we’re going to do precision ag, we should do it right. We do conventional ag pretty well. Why walk away from something that works really well if it’s not giving us something different? Too often, the hard things we just overlook and we’re like, “Look, I’m going to do precision ag, but I’m going to shoehorn this conventional ag into a yield monitor.” The best yield monitor works for conventional ag. But it’s not at the resolution needed for precision ag. The first thing we think about when with precision ag is yield monitoring or maybe guidance. But in its current state, yield monitoring is not precision ag. It’s good enough to kind of give me big, huge coarse trends in yield across big distances.
Luck: It gets back to how big my management zones need to be and until we get to where we can do plant-by-plant control of things, I think that’s probably right, it’s probably about the best we can get.
“If we’re not going to keep doing this applied research, and do it so everybody is comparing apples to apples, how do we bridge that gap?…”
– Joe Luck
McGrath: With precision today, an error gets amplified. You can be further from the right answer when you’re trying to be precise. If you’re doing the average for the whole field, that’s good enough. All of a sudden, when you start going smaller, you’re amplifying the error that we knew we had and it becomes more dangerous. You might be better off doing conventional ag if you’re not getting good, quality data and have a quality plan.
Luck: I see application technology getting to the point where we can do row-by-row control and react within a second of what we need to do. There’s going to be places where it’s going to be economical because there’s got to be other ways of doing it. We can spray so much, so fast until we have either more penalty for doing it that way or more benefit or premium for doing it the other way. It’s probably going to be tough to do.
McGrath: The private sector, while they provide a lot of good machinery and service to that machinery and recommendations, I think there’s cottage industries now exploding for data management, which we need, and data interpretation and inter-operability between sensors. The private industry can specialize and do a very good job, but I think the university level offers the unique opportunity to have psychiatrists or psychologists talk to agronomists about how people make decisions. That’s where that partnership is going to emerge again between the land grant universities and the private sector with the ability to integrate across disciplines and answer “How do we make decisions?”
You need a sociologist to talk about that, and I need an economist, and I need an engineer and I need a mathematician. You’re not just going to have that in the private sector by and large. I think if the innovation is going to occur, we as universities have to help develop that aspect.
