Understanding uncertainty in the carbon footprint of beef production

Greenhouse gas (GHG) accounting models facilitate mitigation of emissions from livestock systems. Such models are approximations, and uncertainties in their output may stem from a) uncertainty or variability in input data, b) uncertainty resulting from model scope and allocation methods, or c) uncertainty in modelling approach used. While sources a) and b) vary depending on the modelled scenario, c), referred to as epistemic uncertainty, relates to the modelling process, and as such is inherent in the methodology used rather than the specific scenario. This study combines a farm-level model comprised of widely used GHG accounting methodologies with a typical northern hemisphere suckler beef production system, and employs Monte Carlo simulation to assess the sensitivity of the modelled GHG footprint to epistemic uncertainty in the model. Following a cradle-to-gate approach, an emissions intensity of 19.20 +/- 2.49 kg CO2-eq kg live weight(-1) was estimated for the modelled system. The study also highlights a discrepancy of 8.3% between deterministically and stochastically calculated emissions; this results from skewness in key modelling coefficients, primarily those relating to nitrous oxide emissions. Sensitivity analysis showed coefficients relating to emissions of nitrous oxide from land and methane from enteric fermentation were most influential in the modelled uncertainty, though coefficients relating to livestock feed production also contributed substantially. In conducting a root-cause analysis of uncertainty in GHG accounting from beef production, this study makes a novel contribution to the literature surrounding uncertainty in livestock emissions modelling. Developers of GHG accounting methodologies may use these insights to focus efforts on refining the most influential elements of these approaches, while researchers applying the models should be aware of the associated uncertainty. The latter should be quantified and effectively communicated where these models are used to support policy decisions. (C) 2019 Elsevier Ltd. All rights reserved.