A groundbreaking study published in Smart Agricultural Technology has revealed that hyperspectral sensors, combined with artificial intelligence, can significantly enhance the measurement of herbicide effectiveness. Researchers from the Arkansas Agricultural Experiment Station, part of the University of Arkansas System Division of Agriculture, have demonstrated that this technology can outperform human evaluations, offering a more accurate and consistent approach to weed control.
The study, led by Aurelie Poncet, an assistant professor in the Department for the Division of Agriculture, and Mario Soto, a master’s student, utilized hyperspectral sensors to measure the response of common lambsquarters to glyphosate, a widely used herbicide. Unlike standard cameras that capture images using visible light bands, hyperspectral sensors record a vast range of bands from 250 to 2,500 nanometers, including thermal infrared. This advanced technology allows for a more detailed and nuanced analysis of plant stress induced by herbicides.
One of the most intriguing findings of the study was the enhancement of photosynthesis in common lambsquarters when exposed to sublethal doses of glyphosate. This discovery not only sheds light on the complex interactions between herbicides and plants but also opens new avenues for research in weed science.
The researchers employed a random forest machine learning technique to analyze the vast amount of data collected during the experiment. This method, which combines the outputs of numerous decision trees to produce a single outcome, achieved a 12.1 percent margin of error in measuring herbicide efficacy. While this is slightly higher than the 10 percent margin of error typically achieved by trained weed scientists, the goal is to refine the technology to surpass human accuracy.
The implications of this research are far-reaching. Hyperspectral sensing has the potential to revolutionize weed management by providing a high-throughput platform for classifying weed responses to herbicides and screening for herbicide resistance. This technology could overcome the limitations of human visual evaluations, which can be affected by fatigue and environmental conditions.
Nilda Roma-Burgos, a professor of weed physiology and molecular biology at the Experiment Station and Bumpers College, highlighted the significance of this development. She noted that while training can compensate for evaluators’ lack of experience, mental and physical fatigue from long workdays can impair judgment, even for the most experienced evaluators. The use of hyperspectral sensing could eliminate the human factor in herbicide efficacy evaluations, providing a more reliable and consistent method.
However, much work remains to validate this method across various weed species, herbicide modes of action, and environmental conditions. The study’s co-authors, including Juan C. Velasquez, Wesley France, Kristofor Brye, Amanda Ashworth, and Cengiz Koparan, represent a collaborative effort to advance this technology.
Funding for the research was provided by the National Science Foundation and the USDA’s National Institute of Food and Agriculture, underscoring the importance of this work in the field of agricultural technology. As this technology continues to develop, it holds the promise of transforming weed science and contributing to more sustainable and effective agricultural practices.