Satellite-based monitoring of forage quality in grasslands of the United Kingdom using Sentinel-2 data and Random Forest Regression

Gonzalo Irisarri^1*Marcos Texeira²Paul Harris³Keith Pembleton⁴

• ¹University of Wyoming, Laramie, United States
• ²Universidad de Buenos Aires Facultad de Agronomia, Buenos Aires, Argentina
• ³Rothamsted Research North Wyke, Okehampton, United Kingdom
• ⁴University of Southern Queensland, Toowoomba, Australia

In the temperate grasslands of the UK, forage quality is a key factor influencing both animal performance and environmental impact. Because forage quality strongly influences rumen fermentation, improving it can reduce enteric methane emissions and mitigate animal nutritional stress. However, large-scale monitoring remains limited due to the reliance on destructive and costly methods. Here, we explored an indicative approach based on optical remote sensing (Sentinel-2) and random forest regression (RFR) models, calibrated with data from near-infrared (NIR) sensors mounted on agricultural machinery, to predict four critical forage quality attributes: crude protein (CP), water-soluble carbohydrates (WSC), neutral detergent fiber (NDF), and acid detergent fiber (ADF). We collected more than 9,500 georeferenced observations between 2020 and 2022 at the North Wyke Farm Platform in southwest UK. These spanned paddocks with permanent and improved pastures. Model performance was strong across all forage quality traits, with R² values ranging from 0.77 to 0.86 and consistently low RMSE values, indicating high predictive accuracy. NIR and red-edge bands were the most influential for prediction. Improved pastures generally exhibited higher forage quality, with lower ADF and higher WSC concentrations than that found in the permanent pastures. Model-predicted seasonal changes were modest, whereas spatial contrasts between paddocks were much stronger. The calibrated models are suitable for forage systems with species composition and quality ranges like those represented in our dataset and should not be directly applied to other forage types. The proposed approach demonstrates the potential of remote sensing for tolerably accurate forage monitoring.