Interpretive Summary: Approaches for predicting dairy cattle methane emissions: from traditional methods to machine learning

By: Stephen Ross, Haiying Wang, Huiru Zheng, Tianhai Yan, Masoud Shirali

This review provides a comprehensive overview of the different modeling approaches taken in the prediction of dairy cattle methane emissions.

Mechanistic models, which mathematically simulate the methane production process of the dairy cattle rumen, are both accurate and adaptable, yet their necessary input data is difficult to obtain and if imprecise, can produce misinformative results.

Empirical models, which statistically quantify the relationships between methane emissions and production factors, are a more accessible alternative to mechanistic models, yet their accessible structure limits them to the same range of data on which they were originally developed.

Machine learning models, which are based on a particular learning pattern, can be trained to identify trends in methane production and use these lessons to make accurate predictions. Their application in the prediction of dairy cattle methane emissions remains scarce, yet those that have been show promising potential.

Commercially deployable models can utilize any of the previous approaches, as long as the traits they use are obtainable in a commercial farm setting. Those developed favor the use of milk fatty acids, yet the variation in their results needs to be consolidated before robust predictions of methane emissions on commercial farms can be achieved.

Read the full article in the Journal of Animal Science.