Montana State University professor Mike Walach flies a drone for a demonstration on Feb. 24, 2023.

Rachel Leathe/Chronicle

Jamie Sherman spends her summers walking barley fields.

She’s the director of the barley breeding program at Montana State University, and she walks the lengthy rows of MSU research farms, looking for the best performing plants. 

Those that are too tall and falling over don’t make the cut. Neither do any infested with disease, or that seed at the wrong time. The breeding program — which analyzes some 50,000 barley lines a year — seeks to fine-tune barley varieties for growers in Montana by breeding them for drought and disease resistance.

Sherman’s decisions are fast and instinctual. Scanning the crop variations with her eyes, Sherman said the process is almost like an art form — one that comes naturally from having done this kind of work for so many years.

But Sherman isn’t alone in making her breeding decisions. A drone overhead has already scanned the same field Sherman has, and it has a special camera to take measurements that humans can’t.

The lab, which has used drones to gather data for the past few years, is comparing drone data to human observations. If the drone measurements prove reliable, they could help speed up the breeding process and reduce lab and labor costs.

The Montana State University barley breeding program used a drone to photograph the variations in a barley field near Bozeman. 

Photo courtesy Jamie Sherman

A drone photo captures variation in a barley field for the Barley Breeding Program at Montana State University. 

Photo courtesy Jamie Sherman

The barley drones are just one use of precision agriculture technology that’s gaining steam at MSU right now.

Using technology to collect data — and then using that data for agricultural decision-making — is a burgeoning field, and it’s helping researchers answer questions they couldn’t before. But there are still economic and technological barriers limiting the widespread adoption of precision agriculture, and some farmers worry machines will replace them. 

Researchers don’t see it that way — if done right, they see it as a way to help them. 

A field of wheat near Rapelje is shown in this 2019 Chronicle file photo.

Rachel Leathe/Chronicle

Sherman’s lab wants to use drone data to help breed barley with deeper root systems, which can access soil water more efficiently and be more successful in drought. To do that, researchers can measure how much water the plant leaves are giving off as a proxy for soil water.

That’s a measurement nearly impossible for a human to do at scale — but for a multispectral camera, it’s a piece of cake.

Scott Powell, an environmental spatial analysis professor at MSU, said multispectral images show how a plant reflects light in ways the human eye can’t see. The images can help researchers answer more difficult questions about plant health.

“Essentially, we’re collecting data across different portions of the electromagnetic spectrum, including the visible bands like red, green and blue that our eyes can see,” Powell said.

“But importantly for crop monitoring, we collect data in the near infrared position of the spectrum, which is non-visible to us, ” he said.

Healthy, green plants follow a certain spectral signature that researchers can see in the imagery — something that changes if a plant becomes stressed.

Using that data, researchers can map the variability of plant health across one field. That variability could be related to moisture limitations, nutrient limitations, or differences in soil chemistry, among other factors, Powell said.

In the past, conventional agriculture treated fields as a single unit, and applied fertilizer and pesticide at one standard rate.

But now, researchers can create mathematical models for input application based on differences in plant health in one field.

That means data can show which sections, for example, have enough nitrogen in the soil — so producers can skip applying fertilizer there and put it in other areas that are under-fertilized.

The same logic goes for other chemicals like herbicide and pesticide. Reducing their use saves producers money and reduces environmental damage.

Bruce Maxwell, an agroecology professor in the land resources and environmental sciences department at MSU, has been experimenting with different rates of nitrogen on farmland for years.

Chuck Merja, a wheat farmer in Cascade County and one of the producers working with Maxwell, said the research has both improved his yields and saved him money.

Chuck Merja, a wheat farmer in Cascade County, stands in his field where Montana State University researchers are testing varying applications of nitrogen fertilizer. 

Photo courtesy Bruce Maxwell

A Montana State researcher flies a drone over wheat farmer Chuck Merja’s field. The research team is testing different rates of nitrogen fertilizer on wheat crops. 

Photo courtesy Bruce Maxwell

By tracking yield information across his fields for several years, Merja could understand where the low and high yielding areas were. His initial goal was to improve the areas with lower yield.

But after several seasons experimenting with different fertilizer rates and seeing no improvement, Merja tried a different approach — moving nitrogen he would have put on low yielding areas to high yielding sections instead.

“There are some areas that are just not going to put out a crop no matter how many resources you throw at them,” Merja said. “Instead we can put that nitrogen on areas that have a proven high performance.”

Several maps show the progression of Bruce Maxwell’s nitrogen fertilizer experiment on a Montana winter wheat field. Researchers collect yield data and decide where to apply nitrogen fertilizer. Then they evaluate the yields and continue experimenting to find optimal fertilizer use. 

Image courtesy Bruce Maxwell

Using resources efficiently is the backbone of precision agriculture. It’s also what Maxwell and other researchers are trying to automate, so that farmers don’t have to figure out the optimal rate and location for inputs themselves.

“So much of precision agriculture has gotten focused on new gadgets that we can add to machines to monitor and measure things, but without thinking about exactly what you’re measuring and why,” Maxwell said.

“If those fancy gadgets don’t help you make better decisions — better meaning, you make more money and reduce pollution — they’re not useful,” he said.

That’s why Maxwell’s team is spearheading technology that can actually help farmers make decisions, rather than just collect data.

Feeding the technology years of a field’s yield information, input rates and weather conditions allows it to make a predictive model of optimal input rates based on what happened in years past. The more data, the more accurate the model.

“The whole thing is automated down to the point where it basically spits out a map that says, here’s what the optimum would be given the data so far,” Maxwell said.

But Maxwell said that sometimes when he presents his research, people bring up concerns that automation will cost farmers their jobs.

It’s important to remember that many farmers have generational knowledge that can’t be replaced by machines, Maxwell said.

“I want to see that traditional knowledge combined with the new approaches,” Maxwell said.

“How people interpret what’s happening on their farm and have for years and years through generations has a lot of value. We should be careful not to lose that.”

Montana State University professor Mike Walach flies a drone for a demonstration on Feb. 24, 2023.

Rachel Leathe/Chronicle

More and more students at MSU are interested in learning about precision agriculture technology.

Some want to gain that expertise so they can integrate tech into their own family farms, said Paul Nugent, an agricultural engineering professor who’s working on creating a precision agriculture minor for MSU.

Nugent and his colleague Jasmine Neupane joined MSU last spring to build the precision agriculture program. This year, they’ve offered four classes covering the basics of the field, and are waiting to hear student feedback before finalizing the minor.

There are still other courses that dial into the specifics of precision agriculture, like one that teaches students how to build and fly their own drones.

Mike Walach, a professor in agriculture technology who focuses on aviation, said in order to do useful work in precision agriculture, people first must learn how to collect data.

Montana State University professor Mike Walach holds one of the drones he uses to teach students on Feb. 24, 2023.

Rachel Leathe/Chronicle

Montana State University professor Mike Walach fires up a drone for a demonstration on Feb. 24, 2023.

Rachel Leathe/Chronicle

The most challenging part of the field is then knowing what to do with that massive amount of data, Walach said.

Processing speed is another obstacle. While much improved from just five years ago, it can still take hours or even days to process the thousands of images taken by drones.

“That’s one of the barriers to a lot of precision agriculture — just dealing with that data quick enough for it to be useful,” Walach said.

And, while the cost of drones and cameras has dropped dramatically in recent years, it’s still expensive.

The high-end consumer and low-end professional drones can cost anywhere between $2,500 and $6,000, depending on its camera and other components, Walach said. Users also typically need a subscription to processing software, which can cost thousands.

“We’re not at the point of huge adoption of a lot of this tech,” Walach said. “I think at the end of the day, the question is — what is it going to do for me? Will I be able to save money? Will it increase my yield? There has to be a benefit. Otherwise, it’s just another gadget.”

Montana State University professor Mike Walach flies a drone for a demonstration on Feb. 24, 2023.

Rachel Leathe/Chronicle

Aside from using drones to monitor crop health, some MSU researchers are experimenting with technology to better understand animal behavior and livestock interactions with predators.

Having a clearer picture of how cattle use landscapes and rangeland helps ranchers manage herds more efficiently, said Tim DelCurto, a professor in range beef cattle and management at MSU.

DelCurto said that new technology is helping researchers answer questions that they couldn’t before — like how grazing patterns are affected by the weather and which animals in a herd eat the most supplemental feed.

“What we’re finding is that a lot of our previous assumptions about animal behavior were wrong,” DelCurto said.

Livestock are equipped with GPS collars that send out a geo-stamped location every five minutes. Researchers are also putting accelerometers into cattle ear tags — essentially a Fitbit for cows that can track their movements and activity, DelCurto said.

That technology can also use cameras to monitor what livestock are eating and even their interactions with predators.

Virtual fencing is another upcoming technology — GPS collars restricted to defined fencing areas could someday replace barbed wire, DelCurto said.

While the upfront cost is high, the price for virtual fencing is comparable to traditional fences over time, because paying for additions or repairs wouldn’t be necessary.

“We could someday be in a situation where a rancher could sit in his office and move cattle to new pasture,” DelCurto said.

Still others are focused on improving existing technology, like cameras that capture predator movement in real time for ranchers.

Since 2020, MSU wildlife biologist Jared Beaver has been working to improve game cameras so that they only capture helpful images. That’s because with current game cameras, 90% of the images are empty with no animals — effectively useless to ranchers, unless they have time to sort through thousands of photos.

But Beaver’s team has created an automated camera that only captures images when a predator is nearby, automatically alerting the rancher in real time so they can respond quickly.

The camera is still in its early stages, and the team is working to improve the automation so it can categorize the type of predator and trigger a response.

To Beaver, the goal of new technology is to improve existing management strategies, not replace them.

“That’s really what drives my work — when there seems to be a need for it,” Beaver said.

“None of this is meant to replace the other tools. Our toolbox is not finite. It’s really a question of, how can we grow that toolbox to make the current tools we have that much more effective.”