Photo by Nikolaos Tziolas, UF/IFAS
Someday soon, farmers might use technology equivalent to Siri or Alexa to check the status of their soil’s quality.
Anastasia Kritharoula, a doctoral student in the Soil Science Artificial Intelligence lab of the University of Florida, works under the supervision of Nikolaos Tziolas, University of Florida Institute of Food and Agricultural Sciences (UF/IFAS) assistant professor at the Southwest Florida Research and Education Center (SWFREC). NASA chose Kritharoula as one of 55 grant award winners across the nation to develop web-based artificial intelligence technology that will tell growers how to improve their soil’s health. The three-year grant will be for a maximum of $150,000.
Kritharoula and Tziolas call their project “Harmonized Remote Monitoring of Soil and Land Dynamics.”
“This project uses smart AI foundation models that learn from different kinds of satellite data all at once,” said Tziolas. “By putting all this information together, we can give growers a clearer picture of their soil and land, helping them make better decisions. We act locally but think globally. Hence, by working with NASA, we can tap into enormous satellite data and build models that help growers here and can be adapted anywhere in the world.”
Because Tziolas and Kritharoula will work in stages, they expect to have early results for pilot areas in Florida and Brazil by 2027. That means farmers could start seeing useful findings well before the project is finished.
“With the growing availability of data from satellites, we have an unprecedented opportunity to improve our understanding of soil and land conditions,” Kritharoula said. “However, challenges like the need for specialized technical knowledge often limit a grower’s ability to use this data.”
To address that challenge, Kritharoula and Tziolas plan to design a foundation model inspired by a concept called “neural plasticity” — the brain’s ability to adapt to new information. The model will integrate data from multiple satellite sources to deliver more accurate and consistent estimates of soil properties such as organic carbon and pH levels.
“Furthermore, we will implement a chat-map system that allows non-expert users to access and interpret soil information easily,” Kritharoula said.
The system will give growers clear, actionable insights without needing special training.
For example, farmers could ask, “What is the soil health status of my field?” or “How have fields degraded over the last five years?” and the platform will give them suggestions.
“It’s not just data … it’s practical guidance they can act on,” Kritharoula said.
Before they make it available to growers, scientists will validate the model on a large dataset, using measurements in two pilot areas, one at and near SWFREC in Immokalee and the other in Brazil.
The technology builds on an earlier version of the chat-based generative AI that Tziolas plans to develop so growers can assess crop damage quickly and accurately after hurricanes.
Source: UF/IFAS
Share this Post