Data processing pipeline for peak alignment and background correction for methane measurements from sniffers installed in automatic milking systems

Accurate measurement of methane concentration from dairy cows is essential for genetic evaluation and effective mitigation strategies. While sniffers installed in automated milking systems (AMS) provide a non-invasive method for capturing enteric methane emissions during milking, attributing these measurements to specific animals remains challenging. Simple timestamp synchronization between AMS and sniffer data is insufficient due to clock drift, cow behavior variability, and background methane interference. In this study, we present a data processing pipeline that enhances the alignment of methane measurements with individual milking events. The proposed pipeline consists of two key components: (1) a peak-detection algorithm that refines the start and end times of methane measurements to improve their attribution to specific milkings, and (2) an estimation of local background methane concentrations using plateau regions within each time series. This two-step approach enhances the accuracy of individual-level methane data and accounts for behavioral variability across cows. By improving the accuracy and reducing variance while maintaining the heritability and repeatability levels of sniffer-derived methane phenotypes, the proposed pipeline provides a practical foundation for large-scale phenotyping and supports the integration of methane traits into genetic evaluation and breeding programs aimed at reducing emissions.