Building the European Environmental Footprint of Food Database

Bridging the data gap in food system sustainability

One of the main goals of the EFF database is to ensure consistency and transparency in environmental footprinting within the food sector. Traditionally, sustainability professionals have often relied on fragmented data sources with varying methodologies, making it difficult to compare the sustainability of different food products or to set reliable benchmarks.

The EFF database addresses these challenges by providing a harmonized PEF-wise footprint database for a wide range of food products commonly consumed in Europe, aligned with the Product Environmental Footprint (PEF) method recommended by the European Commission.

Database scope: from consumption patterns to product selection

To ensure the database reflects what Europeans actually eat, the scoping process began with the EFSA Comprehensive European Food Consumption Database. By aggregating surveys conducted after 2009, our team identified products representative of average EU dietary patterns of adults.

The selection followed a rigorous hierarchical approach using the FoodEx2 system:

• Hierarchy Level 4 served as the starting point, as it is the first level specific enough to define ingredients.

• A 1% cut-off criterion was applied to ensure only relevant products are included; only products accounting for more than 1% of consumption within their group were selected.

• For complex categories, Level 5 products were selected to ensure at least 80% coverage of that category's consumption.

FoodEx2 is a comprehensive standardized system developed by EFSA for the detailed classification and description of food items. It uses a hierarchical coding structure and descriptive facets to ensure consistent data collection across various domains, such as food consumption.

Methodology: a harmonized modeling framework

The EFF database provides results for two distinct system boundaries: cradle-to-supermarket and cradle-to-grave. This dual approach allows for fair comparisons between raw and pre-cooked items while acknowledging that consumer behavior heavily influences the final "grave" stages. For example, the impact between coffee and coffee beans is significantly different due to dilution, and rice is expanded in weight due to absorption of water while cooking.

Modeling principles include:

• Life cycle stages: the system tracks impacts from crop cultivation and animal production through processing, packaging, distribution, and final consumer preparation.

• Allocation rules: to ensure PEF compliance, the database primarily uses economic allocation (e.g., between meat and co-products at a slaughterhouse). Unless the PEFCRs prescribe another allocation type, like biophysical allocation for specific sectors like dairy production systems. 

• Emission modeling: environmental impact as result of emissions, consistently calculated for various life cycle stages, based on specific product PEFCRs or PEF guidance reports. For example: direct and indirect laughing gas emissions are calculated using IPCC (2019) calculation rules for all modelled cultivations.

• The Circular Footprint Formula (CFF): This is the specific PEF way of modelling packaging materials. The CFF is used to accurately model the recycled content and end-of-life (recycling, incineration, or landfill) of packaging materials.

Data selection and the challenge of public access

A unique challenge of the EFF database was the requirement for license-free distribution. To achieve this, the project utilized a strategic hierarchy of data sources:

• Activity data: extracted from international statistics like FAOstat and Eurostat to build specific Life Cycle Inventories (LCIs).

• LCI background data: built using publicly available datasets from ELCD and Agri-footprint.

• Market mixes: the geographical spread of global supply chains was captured by assessing the specific market mix of ingredients for each of the 27 EU countries, using 5-year trade and production averages.

Data generation at scale

Data Generation Pipeline

Developing a database of this size, covering 58.500 unique Life Cycle Inventories (LCIs), requires more than just traditional manual LCA modelling. To handle this complexity, our data team utilized our Data Generation Pipeline (DGP) to enable a systematic, scalable approach to data management and processing.

By using the DGP, we move beyond simple data collection to systematic workflows that ensure:

• Methodological alignment: we use “Life Cycle Engines” to guarantee that modelling rules are applied consistently across all system boundaries.

• Reproducibility: every dataset is built through versioned workflows, creating a transparent and auditable trail from source data to final result.

• Updatability: because the source data is versioned, updates can be implemented consistently across the entire database, ensuring the data does not suffer from methodological drift.

Moving forward: from EFF to the Minerva LCI Database

Looking ahead, we are committed to the continuous evolution of our background data through the Minerva LCI database, the follow-up database to the EFF database. As the core of our LCI expertise, the Minerva database provides PEF-aligned and geographically representative datasets that offer extensive opportunities for further development. 

We are currently transitioning from a static repository toward a dynamic research framework, already delivering the Minerva LCI Database to leading research institutes. By prioritizing methodological innovation and thought leadership, we are defining the next generation of LCA development. 

Stay informed: Read more about our vision for long-term LCA database development in our upcoming article.

Get in touch with our experts