Process-based modelling in farm greenhouse gas assessment reveals site-specific dynamics and limitations of emission factor methods

Abstract Life cycle assessment (LCA) is a well-recognized tool to assess the environmental impact of food production. To assess partial life-cycle greenhouse gas emissions in agricultural systems with variable weather and management, this study integrated process-based agricultural systems modelling (APSIM) with emission-factor-based calculations to develop a system modelling framework. We assessed the greenhouse gas emission intensity of 11 fields from 2016 to 2021 at Boorowa Agricultural Research Station, representing an Australian cropping farm with comprehensive management records. Net greenhouse gas emissions varied widely across fields and seasons, ranging from −3.87 to 6.10 t CO 2 −e ha⁻ 1 . Emissions were not only determined by seasonal climate but also prior-year management decisions, highlighting the need for a system-level perspective. Compared to the LCA-APSIM approach, averaged emission factors tend to overestimate direct N 2 O emission and fail to capture field-scale variability driven by climate and management. This highlights the limitations of the emission factor-based approach. Long-term scenario simulations for a continuous cropping system (canola-wheat-wheat) and a phased pasture-crop system (lucerne (×3)-canola-wheat-wheat, with lucerne ungrazed and 50% cut for hay) clearly demonstrated the trade-off between greenhouse gas emissions and nitrogen input. From a long-term perspective, regardless of management practices or cropping systems, soil organic carbon in agricultural systems will eventually reach equilibrium, after which the system will transition from a carbon sink to a carbon source. To optimize environmental sustainability and food security, advanced farm management strategies must delay the attainment of equilibrium and maximize soil carbon storage potential while maintaining productivity. This study provides new insights into field-scale variability in greenhouse gas emissions, soil organic carbon equilibrium timing, and biases in static N₂O emission methods that have not been quantified in earlier LCA–APSIM applications.