Predicting the number of grazing days from milk Fourier-transform mid-infrared spectral analysis

Abstract

Grazing provides multiple benefits for dairy farming, including reduced feeding costs, improved animal welfare, environmental services, and enhanced milk composition. Current certification schemes often require a minimum number of grazing days, typically verified through grazing calendars, which are time consuming for farmers and costly for industry controls. This study evaluates Ind_Herbage, an indicator derived from bulk milk Fourier-transform mid-infrared spectra to predict the probability of herbage consumption, as a proxy for grazing activity in Wallonia, southern Belgium. The algorithm detects grazing periods by analyzing temporal changes in Ind_Herbage values using a smoothed derivative-based approach with a winter correction to avoid false positives from feed supplementation. Applied to 2 yr of data from 72 Walloon dairy farms (approximately 10,000 milk records per year), the method identified an average of 212 grazing days in 2023 and 221 in 2024, with 97% of farms exceeding the 120-d requirement. The number of grazing periods detected per farm (4–5 on average) was well aligned with regional climatic conditions. These results demonstrate the potential of Ind_Herbage to provide an automatic, large-scale assessment of grazing days, offering a practical alternative to traditional grazing calendars for both farmers and dairy industries.