How CFANS researchers are using AI and satellite data to address global challenges

Artificial intelligence (AI), machine learning and deep learning are rapidly reshaping how researchers understand complex systems — especially when paired with satellite data and remote sensing technologies. By analyzing vast streams of information collected from space and airborne sensors, researchers across the University of Minnesota's College of Food, Agricultural and Natural Resource Sciences (CFANS) are using AI to detect patterns across landscapes, atmospheres and ecosystems at scales that were previously impossible.

These approaches are helping scientists monitor croplands, track air quality and identify invasive species more efficiently and accurately. At the same time, CFANS researchers are asking important questions about when — and how — these powerful tools should be used responsibly.

Here are three examples of CFANS-led research that highlight both the promise and responsibility of AI-driven, remote-sensing science.

Monitoring croplands at scale with machine learning

David Mulla, faculty member in the Department of Soil, Water and Climate, and Philip Pardey, faculty member in the Department of Applied Economics, along with colleagues in the Department of Computer Science are exploring how machine learning, paired with satellite imagery, can transform how croplands are monitored across large regions and long time periods.

Their research shows how automated, AI-driven approaches can map where crops are grown more efficiently and consistently than traditional methods. These advances could improve decision-making related to food security, land use and environmental sustainability, while also identifying technical hurdles that must be addressed before these tools can reach their full potential.

Tracking atmospheric gases from space using advanced algorithms

In a global study led by Kelley Wells, researcher in the Department of Soil, Water and Climate, scientists developed new AI-enabled retrieval methods to analyze more than a decade of satellite data measuring key atmospheric gases. By refining algorithms that interpret infrared satellite observations, the team was able to track volatile organic compounds that influence air quality and climate across the globe. The work reveals how these gases vary by region and over time, and it helps identify gaps in current climate and atmospheric models — supporting more accurate predictions of environmental change.

Mapping invasive species with remote sensing and deep learning 

Ce Yang, faculty member in the Department of Bioproducts and Biosystems Engineering, is principal investigator on a Minnesota Invasive Terrestrial Plants and Pests Center supported project that proposes using deep learning alongside satellite and drone imagery to detect invasive plant species earlier and more accurately.

Focusing on species such as Palmer amaranth, spotted knapweed and Canada thistle, the research aims to replace labor-intensive field scouting with a dual-level remote sensing approach. The goal is to deliver timely, actionable information to farmers, land managers and policymakers — helping slow the spread of invasive species before they cause widespread damage.

Advancing AI with environmental responsibility in mind

As CFANS researchers explore new applications for artificial intelligence, they are also grappling with its broader environmental implications. Nick Phelps, faculty member and head of the Department of Fisheries, Wildlife, and Conservation Biology, said this tension is especially important for researchers working in environmental and natural resource sciences.

“AI is an exciting and powerful tool,” Phelps said. “However, I would encourage people to consider the environmental impacts of AI — really, the data centers that enable it. There is an important and complex tradeoff to the technology that is often overlooked for the sake of progress.”

For many CFANS researchers, Phelps said, the challenge is not only learning how to use AI responsibly, but also deciding when its use is appropriate at all. At the same time, he said CFANS is uniquely positioned to lead in this space.

“CFANS is very well positioned to not only develop new applications for AI technologies,” Phelps said, “but to simultaneously conduct the research necessary to ensure the environmental impacts of AI infrastructure are minimized or mitigated.”

Together, these efforts reflect a CFANS-wide approach to artificial intelligence — one that embraces innovation while remaining grounded in the College’s commitment to sustainability, environmental stewardship and thoughtful science.