Background Agricultural Datasets in LCA Studies: Implications of Database Choice

Agriculture-based products are frequently used as main feedstocks in several sectors (e.g., food, biofuels, consumer goods, etc.).  The selection of background database, with their underlying differences in inventory data and modeling approaches, will have implications for the potential environmental impacts and their interpretation.

LCA practitioners model unit processes and systems based on both foreground and background data. Foreground data is related to the specific processes defined as the main points of interest for the study, while background data consists of both upstream and downstream processes related to the assessed system. In most cases, background data is modeled after external secondary sources, such as commercial LCA databases.

There are multiple commercial LCA databases whose set of listed datasets will depend on the geographical and sector scope of each database. Oftentimes, two or more LCA databases will include the same product system with a similar geographical and technological definition. In practice, LCA practitioners choose between datasets to input background data into their LCA models. This choice is particularly relevant for agricultural crops datasets because it is common for agriculture-based products, regardless of their location in the production chain, to show the upstream agricultural stage as one of their impact hotspots. Hence, the quantification of upstream impacts related to the agricultural stage will be essential to the product’s environmental performance.

What happens during the agricultural stage

Inputs such as fertilizers, pesticides, machines, seeds, and energy carriers are required to cultivate and harvest agricultural crops; these inputs all need to be present on or transported to the farm. In addition, some crops require direct irrigation.

Inputs transported to the farm will influence the crop’s ultimate growth and outputs. Moreover, the reactions between these external inputs and the soils in which the crops are being cultivated will cause diverse emissions to air, water, and soil, called direct field emissions. Fertilizers (whether synthetic or organic) are normally used to supply nutrients to the crops; the main nutrients are nitrogen (N), phosphorus (P), and potassium (K). The application of these nutrients will ultimately support the crop’s development but will also cause several direct field emissions: the use of N-containing fertilizers will cause emissions of ammonia (NH₃), nitrous oxide (N₂O), nitrogen oxides (NO_x), and, in some cases, carbon dioxide (CO₂) because of several volatilization, mineralization, nitrification, denitrification, and dissolution reactions between the applied nutrients and the soil biogeochemistry. Also, the application of fertilizers is closely related to leaching, erosion, and runoff of nitrates (NO₃^-) and phosphates (PO₄^3-), as well as changes in the heavy metal balance of soils. Pesticide application (insecticides, herbicides, and fungicides) will similarly result in the release of toxic active ingredients into the environment. Additionally, the specific farming practices that take place in each system through the use of machinery and its operations will influence certain fertilizer and pesticide-related emissions, as well as demand their own fuel consumption.

How LCA captures the direct field emissions of agricultural systems

The LCA sector has developed robust science-based formulas, calculation procedures, and models to quantify the direct field emissions from a given agricultural system. Each LCA database has defined its main calculation rules and methods. In short, the main principle for these calculations consists of quantifying the net nutrient inputs after the total consumption of either synthetic or organic fertilizers or pesticide composition. Afterwards, each chemical substance's emissions are quantified by following its specific calculation formula, which may also require using parameters related to the physicochemical properties of the cultivation soil (e.g., precipitation, temperature, erosion, clay content, etc.).

Some of the existing main calculation procedures include (but are not limited to):

• Intergovernmental Panel on Climate Change (IPCC) 2006 and 2019 Guidelines for National Greenhouse Gas Inventories
• European Environment Agency (EMEP-EEA) Air Pollutant Emission Inventory Guidebook
• Sustainability Quick Check for Biofuels (SQCB)
• Swiss Agricultural Life Cycle Assessment Method (SALCA)

LCA databases do not use the same field emissions methodologies

Understanding this is crucial for an adequate interpretation of results and selection of background datasets. Each LCA database defines the specific field emissions calculation methods in use.

As an exploratory study, EarthShift Global compared the agricultural cultivation inventories and potential environmental impacts of two crops, Argentinian soybean and Philippine coconut, in two different LCA databases: ecoinvent 3.10 and Agri-footprint 6. To ensure a fair comparison, an economic allocation criterion was selected for the four compared systems in both databases.

This exploratory study had three main phases: 1) Analysis of main formulas for calculation methods, 2) Inventory comparison, and 3) Environmental impacts comparison (through the ReCiPe 2016 midpoint (H) impact assessment method).

The main insights from this study show that:

• Both databases use different calculation methods for nitrate and phosphorus-based emissions. ecoinvent reports the use of the SQCB and SALCA methods, while Agri-footprint uses the IPCC and ReCiPe characterization factors for the same flows. Also, different from the SALCA method, the ReCiPe method does not include phosphate emissions, but only phosphorus emissions.
• There are some differences in the key input data (yield and nutrient intake). Each database has different sources for input data, with, in some cases, a different temporality.
• As a result of the past aspects, directly comparing the soybean and coconut datasets between databases shows considerable differences in most impact categories. Interestingly enough, the results of those eutrophication-related categories are drastically different because of the different field emissions methodologies used by each LCA database.

What can we extract from this?

The complexity of agricultural systems is captured by the LCA methodology through the calculation of direct field emissions, which are a considerable source of impacts. Each LCA database defines its own calculation methods for quantifying these emissions, hence, the methods will not be the same across different LCA databases. This can cause considerable differences in potential impacts when comparing datasets for the same crop between different LCA databases. LCA practitioners must be aware of these differences when choosing datasets, especially for those agriculture-derived products, since the agricultural stage is often identified as a hotspot for impacts.

There are no “better” or “worse” databases or calculation procedures, since they are all based on robust science and methodological developments. In the end, the choice of one dataset over another will ultimately depend on the goal and scope of the LCA study. However, what every LCA practitioner must be aware of are these methodological differences between databases. Mixing datasets from different databases must be minimized or done very carefully, if necessary. Future LCA studies must be mindful of maintaining consistent calculations and comparisons when modeling agriculture-derived products.