A University of Illinois Urbana-Champaign research team has developed a dynamic framework that uses satellite imagery and machine learning to detect tillage practices over large areas and long time periods. The team discusses their methodology and findings in a new paper.

“Conservation tillage helps reduce soil erosion, and it affects soil nutrients and moisture retention. Mapping tillage practices across large areas is also important to quantify soil carbon change. But current data are mainly obtained from farmer surveys, which lack timely and detailed spatial information,” said lead author Xiaocui Wu, a research scientist affiliated with the Agroecosystem Sustainability Center, the Institute for Sustainability, Energy, and Environment, the Center for Advanced Bioenergy and Bioproducts Innovation (CABBI), and the Department of Natural Resources and Environmental Sciences in the College of Agricultural, Consumer and Environmental Sciences at Illinois.

Other studies have applied remote sensing with hyperspectral or multispectral imagery to monitor tillage practices by estimating crop residue. But these approaches are typically limited to smaller regions, and the images are sensitive to soil and weather variations, which can lead to inaccuracies.

“We found that satellite signals could vary across regions, as they are affected by soil type, moisture levels, and weather variables. The model needs to account for those elements,” Wu said.

To address these challenges, the researchers developed a new framework that combines crop residue indices from remote sensing data with environmental factors and machine learning to create a dynamic model. They used the approach to estimate tillage percentage across the U.S. Midwest from 2000 to 2022.

“It is a novel solution as one of the first studies to have this level of detailed, long-term tillage information. We have filled a major data gap and scientific gap for this work,” said Kaiyu Guan, the principal investigator of the study, the founding director of the Agroecosystem Sustainability Center and ACES Levenick Professor. “This is especially valuable for policymakers for conservation planning and policy evaluation.”

Overall, the researchers found that conservation tillage increased gradually across the Midwest for both corn and soybean from 2000 to 2022. The maps also revealed clear differences by crop and region: soybean fields generally showed higher no-till adoption, while corn fields relied more on reduced-till practices, and adoption trends varied substantially across the northern and eastern Midwest.

No-till adoption is more common in drier regions such as the Great Plains, where leaving crop residues on the soil surface helps conserve soil moisture. It is also more prevalent in warmer regions, where slower soil warming under residue cover does not strongly constrain planting, the researchers found.

“Understanding how farmers manage soil is essential for evaluating agriculture’s impacts on soil health, water quality, and long-term resilience,” Guan said. “These insights from our study can help agencies and policy makers refine programs and policies for greater effectiveness in the future.”

The findings are also important for researchers, who implement tillage practice effects in their modeling of soil, water, nutrients, and environmental impacts.

The paper, “A framework to detect tillage practices from space: a demonstration in the US Midwest,” is published in Remote Sensing of Environment [DOI: /10.1016/j.rse.2026.115323].

This work was primarily supported by the USDA National Institute of Food and Agriculture (NIFA) Foundational Program awards, the USDA Risk Management Agency, and the USDA Office of Chief Economist’s Analysis of Sustainable Biofuel Feedstocks project. This work is also supported by NIFA Hatch funding and by a Foundation for Food & Agriculture Research (FFAR) Seeding Solutions Award.