“We Just Call It Work”

Farmers across Louisiana rely on LSU AgCenter’s cutting-edge research and technological advances in what’s now called precision agriculture—using remote sensing, machine learning, and big data on farms—to get bigger yields and larger profits while protecting the environment for future generations. But some just call it work.

 

 

 

Mead Hardwick is a fourth-generation farmer in Tensas Parish in the northeastern part of the state, a couple of miles from the Mississippi River, along Highway 65. He farms with his brother Marshall and his dad, Jay Hardwick. When invited to a Zoom meeting, he joins via video link from his tractor. Why not? The tractor practically drives itself and although Hardwick does have to turn it around when he reaches the end of the field, he’s surrounded by technology to help him and his family optimize every bit of energy and money they spend across Hardwick Planting Company’s 8,000 acres on a portion of Somerset Plantation’s 20,000 acres. The tractor has real-time kinematic (RTK) auto-steer and doesn’t swerve more than an inch to the left or right; there’s a GPS receiver on the roof; and right behind, there’s a variable-rate fertilizer applicator that’s pre-programmed to release specific amounts only in certain areas according to a “prescription map” loaded into the tractor’s on-board computer. On a colorful screen by the steering wheel, Hardwick can watch himself like an avatar advancing through the harvested soybean field where the colors on the map show differences in nutrient levels.
 
Modern farm machinery not only harvests crops but also lots and lots of data. As Hardwick chats about various projects on Zoom, his tractor and other machines can record information about the soil and the crop and link it automatically to longitudes and latitudes. He doesn’t have to walk or drive up and down each row with a clipboard to take notes like “Need more fertilizer on this area about two-thirds of the way down.” The tractor knows exactly where it is and what is needed.
 
“We’ve had a long relationship with LSU and the AgCenter,” Hardwick said. “LSU has helped us find the sweet spot with our variable rate fertility program and how to have truly sustainable agriculture. Working with LSU has also helped me be confident; they’ve answered questions and given us access to tools we otherwise would not have. When we want to test something, it’s ‘Here’s an idea,’ and ‘Let’s go do this.’ Having this relationship with LSU on gathering and analyzing data and really understanding the numbers has been seamless.”

 

“We work hard to make the knowledge tangible; we have to. Otherwise, when we go to a farmer, he’ll say, ‘I can’t apply that, I can’t spend the money,’ and the farmer will then decide to do something else. Just like people talk about evidence-based medicine, we try to do evidence-based agronomy—using actionable data that’s customized for each grower, we can provide what-if scenarios and show exactly how much profit is possible by doing one thing as opposed to another.”—Luciano Shiratsuchi

Hardwick wasn’t destined to become a farmer, nor was his father, Jay Hardwick, who was a college arts professor in photography until he and his wife Mary decided to move back to Louisiana to join her family farm in 1981.
 
“My dad didn’t know a corn plant from a cotton plant when he started,” Mead Hardwick said. “So, he had to latch onto something that quickly could get him up to speed. Luckily, he had LSU and the AgCenter Northeast Research Station, which is only about 13 miles from our farm.”
 
Mead Hardwick himself worked in the finance world and real estate development in Dallas, Texas until 2014.
 
“I should maybe say this off the record, but I actually had an overlap for about a year between farming and real estate because I honestly wasn’t sure if this farming thing was going to pan out,” he said.
 
But it did, and he now looks forward to his own children becoming the fifth generation of the Hardwick Planting Company, which also encompasses forest as a natural wildlife refuge for the Louisiana black bear. On their 8,000 acres of tillable farmland, the family grows soybeans, lots of corn and cotton, and smaller amounts of winter wheat and grain sorghum.
 
When it comes to his and his dad’s relative inexperience with farming in the beginning, Mead Hardwick considers it an asset.
 

“What has allowed us to move into some really advanced things and experiment based on the latest in technology and research with LSU is that we got to break the cycle of ‘Dad did it this way’ or ‘Dad did it that way’ and really take a different approach,” he said. “The younger farming generation like myself sometimes struggle to convince the older generation to move forward, so we’re fortunate. The kind of experimentation we’ve done can’t go on if you’re fighting an intragenerational technology battle. That’s how we got to doing precision agriculture and getting the equipment that has enabled us to embrace it fully.”
 
While Jay Hardwick had no prior experience in agriculture, he quickly got up to speed. He’s now one of the most well-known farmers in the state and was just inducted into the Louisiana Agriculture Hall of Distinction, which recognizes those who have made significant contributions to Louisiana’s farming, ranching, forestry, aquaculture, education, and agribusiness industries. He was recognized for having made cotton farming more environmentally friendly by focusing on production techniques that have minimal impact on the surrounding ecosystem. Jay Hardwick has previously received the National Cotton Achievement Award, the National Award for Environmental Sustainability, and the U.S. Environmental Protection Agency’s Environmental Excellence Award. He also serves as chairman of the American Cotton Producers of the National Cotton Council.
 
The Hardwicks work in close collaboration with Precision Ag Coordinator Luciano Shiratsuchi, associate professor in the LSU School of Plant, Environmental & Soil Sciences, and Thanos Gentimis, assistant professor in the LSU AgCenter Department of Experimental Statistics. While Shiratsuchi is an agronomist, Gentimis is not—he’s a mathematician who also happens to be quite good at chess. LSU AgCenter hired both of them to help launch the university’s digital agriculture program in 2018 and interest among students has skyrocketed. Gentimis came from health informatics and now is applying the same expertise (machine learning, artificial intelligence) to improving agriculture, so it’s not lost on him nor his students that the toolboxes they’re building could be valuable in many contexts. A recent graduate got a job as a data analyst to help improve the safety of Louisiana buildings.

 

By using machine learning and artificial intelligence, Gentimis is developing software that can learn from outcomes on one farm or with one particular crop or variety to make predictions for others. It’s all about pattern recognition.

Using vast datasets, Gentimis trained a computer system to recognize the edges of fields and areas with different amounts of vegetation based on drone and satellite images (remote sensing). Next, each defined area is analyzed for plant vigor—the normalized difference vegetation index, or NDVI, can measure plant health and capture information not yet visible to the human eye—pinpointing spots that might need more seed or fertilizer and getting that information to the farmer in real time. The biggest advantage Hardwick sees in this approach is its incredible speed, in two ways. First, he can buy a satellite image of their vast farm, use it to generate a prescription map, and have the tractor out rolling in the field the next morning, applying what’s necessary. Second, the research findings from Shiratsuchi and Gentimis are already at scale—there’s no need to translate or extrapolate from research results in a lab or on a small test patch to the 8,000-acre farm.
 
“Anytime somebody comes up with an idea in a lab or a small plot, you say, ‘Those are very positive results, but can that be replicated on a large scale?’,” Hardwick said. “That’s not to say research isn’t always useful—it gives you an idea of what’s possible. But it often has a ways to go before you can translate it on the farm and it becomes economically viable. Also, what we’re trying to figure out is what we can do today—not tomorrow, but today. With the tools I have in my toolbox, what will that let me do today?”
 
Here, he speaks the same language as Shiratsuchi.
 
“We work hard to make the knowledge tangible; we have to,” Shiratsuchi said. “Otherwise, when we go to a farmer, he’ll say, ‘I can’t apply that, I can’t spend the money,’ and the farmer will then decide to do something else. Just like people talk about evidence-based medicine, we try to do evidence-based agronomy—using actionable data that’s customized for each grower, we can provide what-if scenarios and show exactly how much profit is possible by doing one thing as opposed to another.”
 

The amount of data processing necessary, however, cannot be done on a regular computer. Gone are the days of “let me check the books” or transferring years of farm data on a flash drive.

“The massive tangle of raw data that comes from precision agriculture is like a fertile field full of potential but, just like in farming, if the right tools and seeds are not used, the field will never produce much,” Gentimis said. ”We believe the right tools can be found in data analytics and  machine learning, since the datasets involved have long surpassed the abilities of traditional models for analysis and prediction.”

To meet the research program’s needs for high-performance computing, Gentimis combines local servers with resources in the cloud. He and Shiratsuchi also collaborate closely with industry. Ag-Analytics, a farm management platform led by former Cornell University associate professor Josh Woodard has quickly developed a global reach in helping farmers improve efficiency by melding automated sensor data from farm equipment with satellite imaging and machine learning. When the Ag-Analytics team launched their university partnerships, they came to LSU first.
 
“We went to where the top researchers were, and it’s become a great partnership,” Woodard said. “Farmers on our platform have an option to confidentially share aggregated datasets with researchers at their land-grant university. Our system translates and calibrates the data so we can layer it all onto one cohesive geospatial grid to make predictions in real time to help farmers and support the land-grant extension system.”
 
The crops Shiratsuchi and Gentimis have focused on optimizing so far are sugarcane, soybeans, rice, corn, cotton, sweet potatoes, wheat, and sorghum—industries with an estimated total annual economic impact of about $6.4 billion.
 
“Short of launching hybrids that don’t need fertilizer, precision agriculture is the only way for farmers to save money and the environment, using fewer resources while getting the same or better yield,” Shiratsuchi said. It’s the only practical way to improve efficiency and reduce contamination, using only what you need and where you need it.”
 
Mead Hardwick agrees.
 
“We grow commodity crops for which there is a set market price,” he said. “We can’t do much about the price, so to increase our profitability, we need to control our cost, produce more, and be more efficient. Being strategic with our nutrients, especially, has been key for us. It feels good, too, considering how nutrient runoff contributes to the ‘dead zone’ in the Gulf of Mexico.”
 
When the Mississippi River meets the Gulf of Mexico, it brings agricultural runoff from about a third of the country with it, as far north as Minnesota. When fertilizer, such as nitrogen and phosphorus, hits the Gulf, it’s a feast for algae, which multiply. But as the algae die and decompose, they consume the oxygen in the water, causing “dead zones” void of aquatic life. This low-oxygen condition is known as hypoxia, and the Hardwicks and the LSU AgCenter are currently involved in a four-year research project funded by the Patrick F. Taylor Foundation to help solve this problem.
 
“Overall, in America, we’re still applying much more nitrogen than we need,” Shiratsuchi said. “But you don’t gain anything from applying the same blanket amount across the board; instead, you lose. You can’t treat the farm as a uniform field and use averages for everything.”
 
By using machine learning and artificial intelligence, Gentimis is developing software that can learn from outcomes on one farm or with one particular crop or variety to make predictions for others. It’s all about pattern recognition. And while we’re still far from “self-driving farms,” the benefit of data-driven “smart farms” is becoming clearer each day—especially as rapidly changing environments make it harder and harder to rely on experience.
 
Even when something goes terribly wrong, precision agriculture technology can help. When Hurricane Laura wreaked havoc in southwestern Louisiana this August, farmers and insurance agents were able to assess and document the damage faster using remote sensing, so insurance payments could go out.
 
Advanced data analytics will increasingly become part of any “normal” day on the farm, researchers and farmers predict.
 
“I love tech, but I don’t think of it as ‘tech’ or ‘precision agriculture,’” Hardwick said. “We just call it work.”

 

 

 

Elsa Hahne
LSU Office of Research & Economic Development
ehahne@lsu.edu