A multilevel carbon and water footprint dataset of food commodities

Abstract

Informing and engaging citizens to adopt sustainable diets is a key strategy for reducing global environmental impacts of the agricultural and food sectors. In this respect, the first requisite to support citizens and actors of the food sector is to provide them a publicly available, reliable and ready to use synthesis of environmental pressures associated to food commodities. Here we introduce the SU-EATABLE LIFE database, a multilevel database of carbon (CF) and water (WF) footprint values of food commodities, based on a standardized methodology to extract information and assign optimal footprint values and uncertainties to food items, starting from peer-reviewed articles and grey literature. The database and its innovative methodological framework for uncertainty treatment and data quality assurance provides a solid basis for evaluating the impact of dietary shifts on global environmental policies, including climate mitigation through greenhouse gas emission reductions. The database ensures repeatability and further expansion, providing a reliable science-based tool for managers and researcher in the food sector.

Fig. 1

General outline of the different construction steps of SU-EATABLE LIFE (SEL) database. Step 1 includes the preparatory phase where studies were collected from literature and public repositories, selected on the basis of eligibility criteria, CF and WF values of food commodities extracted and harmonized, and then reported into the level 1 information of the SEL database. In Step 2 the other layers of information are created which represent different levels of aggregation of data reported in level 1. CF and WF values statistical analysis are reported for food items (level 2), typologies (level 3) and sub-typologies (level 4). In Step 3 the complex set of data reported in Level, 1, 2 and 3 are summarized into an easy to use dataset suitable for quick consultation by technical and not technical users.

Fig. 2

Method for attribution of CF (or WF) value to a food item based on data quality flags. The scheme shows the procedure applied to evaluate the level of uncertainty associated to CF or WF value of a food item and how this information is used to decide the best value that should be used to represent the item. Three quality flags related to a statistical aspect of the data population are calculated to attribute the level of uncertainty. Each flag has different level of quality, red being the worst, green the best. Flags are then combined and expert judgement is used to associate a suggestion for data use to each flag combination. If the item median value is characterized by high uncertainty it poorly represents the item and caution is needed to use this data to represent the food commodity, the users is therefore redirected to a higher level of aggregation such as the sub-typology or the typology which includes the analysed item.

Fig. 3

CF value of vegetables vs. their yield. Carbon footprint value of food items included in the typology “vegetables outdoor” is plotted versus their average yield value as reported in FAOSTAT (data EU-28, year 2017).

Fig. 4

Kernel (Gaussian, bandwidth 0.0433) density estimate (a) and box-plot (b) of potato CF data. Kernel (Gaussian, bandwidth 0.3325) density estimate (c) and box-plot (d) of maize CF data. Empirical distribution for CF data of potato and maize as reported in studies listed in Level 1 of the database (data analysis done with STATA 14.2).