The role of uncertainty in representative Italian LCA database: the case of olive datasets

Data uncertainty plays a crucial role in influencing the quality of results obtained from a Life Cycle Assessment (LCA) study. Alongside, the use of representative data for the agrifood sector is fundamental, on the one hand, to account for the various technological, biological, and environmental factors affecting its related productions and, on the other, to reduce their related uncertainty. In this regard, as part of a research project (PRIN 2017, ID code 2017EC9WF2) funded by the Ministry of University and Research (MUR), the Italian Life Cycle Inventory Database of Agrifoods (ILCIDAF) database has been developed in order to provide LCA practitioners with regionalised data for four mains Italian agrifood products, i.e., olive oil, wine, citrus and bread/pasta. This paper aims to analyse two different approaches to calculate the uncertainty related to the regionalised data for Italian olive production included in the ILCIDAF database. In addition, these two approaches are compared in order to understand how these data and their related uncertainty may affect the final results. In the ILCIDAF database, the datasets for olive production have been developed by normalising the input and output to the olive yield of 19 Italian regions and considering its temporal variation between 2015 and 2020. In this context, the first approach consists of calculating the uncertainty connected to the temporal fluctuation of each region, assuming that the uncertainty is related to the annual variation of olive yield among regions. Instead, the second approach involves the use of the basic uncertainty reported in Ecoinvent for some categories of products and emissions. Results underscore that the uncertainty calculated according to the annual olive yield fluctuation is overestimated for the input and underestimated for the direct emissions when compared to the one based on Ecoinvent. Consequently, this contributes to differences in the uncertainty of the LCA outcomes, highlighting that the use of inappropriate uncertainty values may significantly affect the results.