Skip to main content
ARS Home » Plains Area » Mandan, North Dakota » Northern Great Plains Research Laboratory » Research » Publications at this Location » Publication #422120
Research Project: Transdisciplinary Research that Improves the Productivity and Sustainability of Northern Great Plains Agroecosystems and the Well-Being of the Communities They Serve

Location: Northern Great Plains Research Laboratory

Title: Evaluating remote sensing resolutions and machine learning methods for biomass yield prediction in Northern Great Plains pastures

Author
item SUBHASHREE, SRINIVASAGAN - Cornell University
item IGATHINATHANE, CANNAYEN - North Dakota State University
item Hendrickson, John
item Archer, David
item Liebig, Mark
item Halvorson, Jonathan
item Kronberg, Scott
item Toledo, David
item SEDIVEC, KEVIN - North Dakota State University

Submitted to: Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/24/2025
Publication Date: 2/26/2025
Citation: Subhashree, S.N., Igathinathane, C., Hendrickson, J.R., Archer, D.W., Liebig, M.A., Halvorson, J.J., Kronberg, S.L., Toledo, D.N., Sedivec, K. 2025. Evaluating remote sensing resolutions and machine learning methods for biomass yield prediction in Northern Great Plains pastures. Agriculture. 15(5). Article 505. https://doi.org/10.3390/agriculture15050505.
DOI: https://doi.org/10.3390/agriculture15050505

Interpretive Summary: Predicting forage yield, the amount of plants available for haying or for animals to graze, is important but time consuming. Using remote sensing, measurements collected by satellites, can speed up this process. Remote sensing can also allow the information to be collected for a larger area. Unfortunately, accurately predicting forage yield using remote sensing has been difficult. Machine learning is useful for analyzing complex interactions. It has shown promise in agricultural uses such as crop yield predictions and nutrient management. This study used hand-collected forage yield data to develop yield predictions from satellite and weather data The combination of satellite and weather data resulted in 52 different features that could be analyzed. Using machine learning to build these prediction models was successful. The study suggested that finer satellite imagery such as CubeSat was better in predicting forage yield. Results also showed the model could be used for high value crops such as alfalfa. This process could be used to make real-time forage yield predictions for use by livestock producers and land managers.

Technical Abstract: Predicting forage biomass yield is critical in managing livestock since it impacts livestock stocking rates, hay procurement, and livestock marketing strategies. Only a few biomass yield prediction studies on pasture and rangeland exist despite the need. Therefore, this study focused on developing a biomass yield prediction methodology through remote sensing satellite imagery (multispectral bands) and climate data, employing open-source software technologies. Biomass ground truth data were obtained from local pastures. Remote sensing data included spatial bands (6), vegetation indices (30), and climate data (16). The top-ranked features (52 tested) from recursive feature elimination (RFE) were short wave infrared 2, normalized difference moisture index, and average turf soil temperature in the machine learning (ML) model developed. The random forest (RF) model produced the highest accuracy (R2 = 0.83) among others tested for biomass yield prediction. Applications of the developed methodology revealed that (i) the methodology applies to other unseen pastures (R2 = 0.79), (ii) finer satellite spatial resolution (e.g., CubeSat; 3m) better-predicted pasture biomass, and (iii) the methodology successfully extended to high-value alfalfa hay crop with excellent yield prediction accuracy (R2 = 0.96). The developed methodology of RFE for feature selection and RF for biomass yield modeling is recommended for biomass and hay forage yield prediction.