Artificial intelligence can be a drain on the environment but, in some scenarios, it may also help solve one of the Chesapeake Bay’s biggest problems: over-fertilizing crops.
An AI program, PlantMap3D, will help farmers apply only as much fertilizer as they need to supplement nutrients left behind by cover crops.
Cover crops, usually grasses and legumes, are grown by farmers during the offseason when fields aren’t producing their main cash crops. Cover crops reduce nutrient pollution in agricultural runoff by reducing erosion and taking up nutrients, storing them in the plants and the soil for later use by cash crops.
The AI technology will analyze the cover crops and calculate the amount of nutrients the plants add to the soil before the farmer plants the cash crop. Environmentalists hope it will reduce nutrient pollution in waterways, and it could also demonstrate the limitations of cover crop nutrients. Enrolled farmers in three Bay watershed states will start using the program this spring.
In excess amounts, nutrients from chemical and manure fertilizer can create “dead zones” in the Bay that deprive aquatic species of oxygen. Based on a scientific model from the Chesapeake Bay Program, conservation practices like cover crops have reduced nutrient pollution in the Bay. The model estimates that the two main forms of nutrients, nitrogen and phosphorous, entering the Bay have decreased by 15% and 22% respectively between 2009 and 2024. But it’s unclear if that’s truly what’s happening on the ground.
Mike Twining, vice president of innovation at Willard Agri-Service, which helps implement the AI tool on farms, said that while not all Bay goals have been met, “I also would [say] we’ve come a really long way.” He added that “technologies like this will enable us to get better.”
The Nature Conservancy, in partnership with Willard Agri-Service, Growmark FS, North Carolina State University and the U.S. Department of Agriculture, plans to bring the AI program to 150,000 acres across Maryland, Delaware and Pennsylvania.
Backed by about $16 million from the USDA’s Natural Resources Conservation Service and about $11 million in partner contributions, the tool will be free to eligible farmers. And if they end up seeing a dip in production, the program will pay them up to $50 per acre.
Different cover crops work with nitrogen in different ways. Legumes, such as peas, are nitrogen “fixing” crops because nitrogen gas in the plant’s roots is converted to ammonia by soil bacteria. Ammonia is a form of nitrogen that the plants readily consume.
Grasses like rye and oats, by contrast, are nitrogen “scavengers.” They tend to absorb nitrogen from the soil — though they return it to the soil after the cover crops decompose on site.
Many farmers use a mix of cover crops because one species can enhance the other. For instance, grass and legume cover crops can even out each other’s rates of decomposition to give time for microbes to eat up the nutrients. But sometimes one crop can overpower the other, creating patchy spots and an unclear picture of how much nitrogen is present.
“Cover crops are a great tool and good practice, but they bring complexity,” said Chris Reberg-Horton, North Carolina State University professor.
This is where PlantMap3D comes in. It starts with cameras on farm equipment that take pictures of the cover crops as farmers drive the equipment across the field. Then, the AI software will identify the plant species in all the photos. The process produces a map and measures how much biomass of each species is present.
Simultaneously, Reberg-Horton feeds the model with information about the chemical makeup of the soil, recent weather patterns, the plant species and their growth stages. By the end, the farmer has a map showing which areas of their land need more or less fertilizer.
The Nature Conservancy expects this program will prevent three million pounds of nitrogen from being applied to fields and potentially running off into Bay waterways. According to Reberg-Horton, the amount of fertilizer applied after using the technology varied at experimental farms.
Twining, with Willard Agri-Service, said the program will help farmers understand the biology of cover crops. “This deeper understanding” he said, “will ultimately create the win-win everyone is looking for — better environmental outcomes and more profitable Mid-Atlantic farmers.”
Charlie White, professor of nutrient management at Penn State University, said PlantMap3D might not provide a farmer with the full picture of their nitrogen fertilizer needs because it only provides nutrient information on cover crops, without accounting for nitrogen that may be available from the soil as well. PlantMap3D relies on the farmer to calculate how much fertilizer they’d use if they didn’t have the cover crop information as a starting point.
At the very least, White said, it could help farmers understand the short-term nutrient impact of their cover crops.
“We need lots of tools available,” White said. “I’m glad that [PlantMap3D] is getting people to talk about the need for improving nitrogen fertilizer recommendations. All of those things are critically needed and helpful.”