Evaluation of models to predict methane emissions from enteric fermentation in North American dairy cattle

E. Kebreab, J. France, B.W. McBride, N. Odongo, A. Bannink, J.A.N. Mills, J. Dijkstra

Research output: Chapter in Book/Report/Conference proceedingChapter

20 Citations (Scopus)

Abstract

The increasing focus on the environmental impact of agriculture means that there is a need for an accurate inventory of greenhouse gases (GHGs) to identify main sources of pollution and evaluate effects of potential mitigating options. A number of models have been developed to predict methane emissions from enteric fermentation applicable to North American cattle. The objectives of this work are to collate data on methane emissions from the literature and evaluate selected models using these independent data. Six models are considered: linear model of Moe and Tyrrell (1979b), linear and non-linear models of Mills et al. (2003), dynamic model of Kebreab et al. (2004) and Tier I and Tier II models recommended by the Intergovernmental Panel on Climate Change (IPCC, 1997). A database consisting of 47 records, not used in the development of the models, was used. Analyses were done on lactating cows only and on both dry and lactating cows. Assessment of the error of prediction relative to observed values was made by calculation of mean squared prediction error (MSPE). In addition, concordance correlation coefficient (CCC) was also used to evaluate if predicted values were precise and accurate when compared to observations. When data from lactating cows only were used, the models tended to underpredict methane production, except for the linear model of Mills et al. (2003). The linear model of Mills et al. (2003) predicted methane production from lactating dairy cows better than from a mixture of dry and lactating cows, while the opposite was true for the other linear model. The non-linear model improved MSPE and random error considerably accounting for more than 98% of MSPE for both data-sets. The dynamic simulation model also gave accurate and precise prediction for both sets of data with bias correction factor close to unity. Tier I model underpredicted mean methane production by 4%. Tier II model performed as well as the other linear models for the mixed data-set but not for the lactating cows data-set. The linear models are recommended for use if there is a lack of nutrient information and within the range in which they were developed. The non-linear model can be used for extrapolation but for assessment of mitigation options, more mechanistic models are recommended. Tier I model may be adequate for general inventory of methane emissions but Tier II model requires further refinement.
Original languageEnglish
Title of host publicationNutrient Digestion and Utilization in Farm Animals: Modelling Approaches
EditorsE. Kebreab, J. Dijkstra, A. Bannink, W.J.J. Gerrits, J. France
Place of PublicationWallingford
PublisherCAB International
Chapter27
Pages299-313
Number of pages15
ISBN (Print)9781845930059
DOIs
Publication statusPublished - 2006

Fingerprint Dive into the research topics of 'Evaluation of models to predict methane emissions from enteric fermentation in North American dairy cattle'. Together they form a unique fingerprint.

Cite this