In a recent study published in Smart Agricultural Technology, researchers from the Arkansas Agricultural Experiment Station, the research arm of the University of Arkansas System Division of Agriculture, demonstrated that hyperspectral sensors, such as a spectroradiometer, can assist in measuring herbicide effectiveness, a crucial component of weed control that reduces herbicide resistance.
HYPERSPECTRAL — Mario Soto, left, and Aurelie Poncet, demonstrate a hyperspectral sensor used in a study to quantify herbicide effectiveness on plants. Image Credit: Paden Johnson | U of A System Division of Agriculture
Researchers in Arkansas have created a system that measures herbicide-induced plant stress more accurately than humans can by fusing artificial intelligence with sensors that can see beyond visible light.
Hyperspectral sensing records bands from 250 to 2,500 nanometers and thermal infrared, whereas standard cameras employ the three visible light bands—red, green, and blue—to produce images in the spectral range of 380 to 750 nanometers.
The researchers utilized this method to determine how common lambsquarters responded to glyphosate. They also discovered empirical evidence that photosynthesis in the plant was enhanced when exposed to a sublethal herbicide dosage. Common lambsquarters (Chenopodium album L.) is a weed in agriculture and gardens.
Plant response to herbicide application is measured using visual ratings, but accuracy varies with the quality of training and years of practice of the rater. We thought, if we could have a sensor that automates some of this decision, we might be able to implement it into applications down the road.
Aurelie Poncet, Study Principal Investigator and Assistant Professor, Department for the Division of Agriculture, Dale Bumpers College of Agricultural, Food and Life Sciences
Weed scientists are trained to assess herbicide efficacy with a 10% margin of error, plus or minus 5%. The researchers used machine learning models on spectroradiometer data to achieve a 12.1 percent margin of error. Their objective is to reach below ten percent.
The researchers employed a random forest machine learning technique to examine the millions of vegetation index data points obtained throughout the experiment. The method combines the outputs of numerous decision trees to produce a single outcome.
Our success using random forest to describe common lambsquarters response to glyphosate application opens the possibility of moving beyond the development of vegetation indices, another approach gaining traction in the published literature.
Mario Soto, Study Lead Author and Master’s Student, Dale Bumpers College of Agricultural, Food and Life Sciences
Next Steps
Once developed, hyperspectral sensing might be utilized to monitor individual weed responses to herbicide treatment while overcoming the constraints of a human visual evaluation. According to the study's authors, further development and validation of the technology might create a platform for high-throughput classification of weed response to herbicides and herbicide resistance screening.
While training can compensate for evaluators' lack of experience, mental and physical fatigue from long workdays evaluating treatments in harsh environmental conditions can impair judgment for even the most experienced evaluator, according to Nilda Roma-Burgos, professor of weed physiology and molecular biology at the Experiment Station and Bumpers College.
This method, in principle, could remove the human factor in herbicide efficacy evaluations and will be an invaluable research tool for weed science. Meanwhile, much work still awaits to validate the method across key weed species, herbicide modes of action, time after herbicide application and environmental conditions.
Nilda Roma-Burgos, Study Co-Author and Professor, Weed Physiology and Molecular Biology, Dale Bumpers College of Agricultural, Food and Life Sciences
The study's co-authors were Juan C. Velasquez, a graduate research assistant in weed science of the crop, soil and environmental sciences department; Wesley France, a program associate; and Kristofor Brye, a professor of applied soil physics and pedology at the university.
Co-authors included Amanda Ashworth, a research soil scientist with the Agricultural Research Service of the U.S. Department of Agriculture, and Cengiz Koparan, an assistant professor of precision agriculture technology in the departments of biological and agricultural engineering and agricultural education, communication, and technology.
The research on hyperspectral imaging received partial funding from the National Science Foundation’s NSF-SBIR Phase II Award No. 2304528 and the USDA’s National Institute of Food and Agriculture.
Journal Reference:
Soto, M., et al. (2025) Hyperspectral indicators and characterization of glyphosate-induced stress in common lambsquarters (Chenopodium album L). Smart Agricultural Technology. doi.org/10.1016/j.atech.2025.100890.