A parsimonious software sensor for estimating the individual dynamic pattern of methane emissions from cattle

Area of Science:

• Agricultural Science
• Environmental Science
• Animal Science

Background:

• Estimating methane emissions from cattle is vital for environmental monitoring and livestock management.
• Accurate and cost-effective methane predictors are needed for large-scale livestock farming applications.

Purpose of the Study:

• To integrate real-time animal feeding behavior data with an in silico model for predicting individual methane emissions in cattle.
• To develop a software sensor for dynamic methane emission prediction within a precision farming context.

Main Methods:

• Developed a dynamic parsimonious grey-box model using ordinary differential equations.
• Utilized dry matter intake (DMI) or intake time (IT) as predictor variables.
• Model building was supported by experimental methane emission data from respiration chambers in finishing beef steers.

Main Results:

• The software sensor, operating off-line, analyzed 37 individual methane production patterns.
• Model predictors DMI and IT showed similar performance, with an average Lin's concordance correlation coefficient (CCC) of 0.78.
• Daily methane production prediction achieved a CCC of 0.99 for both DMI and IT predictors, indicating excellent model performance.

Conclusions:

• The developed in silico model, integrated with a software sensor, performs well in predicting methane output.
• Intake time (IT) measurements are easier to obtain than dry matter intake (DMI), suggesting IT-based sensors are a viable solution.
• This approach offers a promising strategy for large-scale, accurate methane emission prediction in cattle.