Challenges in using soil carbon modelling in LCA of agricultural products—the devil is in the detail

Purpose Currently, there is no consensus on how the impacts of land use on the soil organic carbon (SOC) stocks would be best quantified within life cycle assessments (LCA) of agricultural products. The impacts of different decisions were tested within a model-based assessment of soil carbon changes on the life cycle global warming impact for spring wheat produced in two example regions in Finland (Southwest Finland and Northern Savonia) on mineral fields. Methods Global warming impact for spring wheat was assessed, including CO2 emissions due to the SOC change. The SOC change assessment was made with the soil carbon model Yasso07. The effects of assumptions on land use history were tested, i.e. the initialisation of the model and time horizon of the analysis (20 or 100 years) on the SOC change estimates. Other greenhouse gas emissions contributing to the global warming impact of spring wheat production were assessed using general LCA methodology taking into account the greenhouse gas emissions caused by the production of input materials and fuels, as well as direct and indirect N2O emissions from the soil due to fertilising and the decomposition of crop residues and organic matter, nitrogen leaching and volatilisation and lime application. Results and discussion The selection of the model initialisation method and timeframe remarkably affected the SOC change estimates. The global warming impact of wheat production, without accounting for SOC changes, was 0.68 and 0.89 kg CO2-eq/kg yield in Southwest Finland and Northern Savonia, respectively. The impact of SOC stock changes on the total global warming impact varied from –4 to 5% in Southwest Finland and from 5 to 21% in Northern Savonia, depending on the assumptions used to initialise the model or the timeframe applied in the analysis. Adding a cover crop as a means to increase the SOC stock removed between –67 and –26% of the total global warming impact in both regions. Conclusions It is essential that all the decisions made in the analysis are transparently reported and communicated. The choice of assumptions regarding the reference state, model initialisation and time horizon of the assessment period should be made based on the scope and goal definition of the LCA study.