How does LandGriffon measure environmental impacts?

LandGriffon helps companies strategize the sustainable transformation of their supply chains by using technology, data and scientific-based analysis to manage environmental impacts. This blog is part of our methodology series. You can find more blogs on our methodology here, or head to our website and download the complete paper.

Calculating impacts

Version 0.1 of LandGriffon includes baseline indicators of environmental impacts such as water use, land use, deforestation, greenhouse gas emissions, and biodiversity loss associated with agricultural production.

Once a company’s data is geolocated, we compute indicators, denoted by the symbol I, of the materials sourced. All indicators are calculated as the quantity of a commodity sourced multiplied by an impact factor, which represents the average impact per ton of the commodity produced across the sourcing geometry. So, the impacts, Icg, associated with commodity, c, and geometry, g, can be calculated as:

(Eq. 1)

where IFc,g is the Impact Factor for the commodity, c, across sourcing geometry, g, in impact per ton of commodity produced; and Sc,g is the total quantity of commodity c sourced from geometry g in tons.

Impact indicators calculation depends on available data, which varies across impacts. For example, spatial maps of crop production exist, as do maps of water use and deforestation. However, spatially explicit information on greenhouse gas (GHG) emissions from agricultural commodity production is not generally available. So, impact calculation of GHG emissions will rely on life cycle assessment or footprinting impact factors.

National and administrative data

For indicators derived from national or administrative-level data (e.g. from generic life cycle assessment or footprint analysis), the closest matching impact factor for the material and administrative region is identified:

• Materials are matched using an extension of the World Customs Organization’s hierarchical Harmonized System (HS) commodity codes. Where there is no exact match, the closest parent in the materials and administrative regions hierarchy is used. E.g. if the impact factor table does not include a record for a given country it will use a global average impact factor.
• Matches in the material hierarchy are selected over matches in the administrative hierarchy. E.g. for organic cotton from Burkina Faso, a global impact factor for organic cotton will be preferred to a Burkina Faso-specific impact factor for generic cotton.

Spatially explicit data

For indicators derived from spatial datasets, the method used to derive the impact factor depends on the location type and whether the indicator is a farm-level or landscape-level indicator.

• Farm-level indicators aim to capture those impacts that occur within the current footprint of agricultural production.
• Landscape-level indicators represent the potential impact of commodities at the landscape-level including areas surrounding production areas under the assumption that commodity-driven land cover change occurs in the proximity of existing farms. At a macro-scale, using land for one commodity sourced into a supply chain presents a land opportunity cost in relation to using land for the production of another crop. Land area sourced into a company’s supply chain adds to the cumulative pressure for land in the proximal area and so contributes indirectly to land use change there.

Materials are assumed to be sourced in proportion to the amount of production in any given location, such that areas that produce more are counted more heavily. For each material and location, we compute a production weighted average impact factor, defined as the sum of the spatial dataset multiplied by the production for each grid location within the sourcing geometry, divided by the sum of the production over the entire sourcing geometry:

(Eq. 2)

Where IFc,i represents the spatially varying impact over each pixel or other spatial unit i of the n such units within the sourcing geometry g, and, Pc,i, is the production of the commodity c across those spatial units (tons). The production weighted average impact factor is a close analogy to the Commodity Supply Mix developed for Life Cycle Assessment (Lathuillière et al. 2021).

Figure 1. (Top) Illustrative description of the weighted average impact factor calculation procedure for forest loss for palm oil in a sub-national region. (Bottom) Detailed description of the weighted average impact factor calculation for a given commodity and geolocation.

Farm-level impacts

Farm-level impact indicators (e.g. Figure 2) are calculated as follows:

• The impact factor is calculated as the production weighted average of the spatial dataset within the sourcing geometry.
• If there is no overlap between the production map and the sourcing geometry, the impact factor is calculated as the area average for the sourcing geometry.
• For farm-level impacts associated with supplier aggregation points, a buffer is applied around the aggregation point. In version 0.1 of LandGriffon this buffer defaults to 50km radius, as is commonly used for palm oil mills.

Figure 2. Example of a farm level impact calculation. Calculation of the land use impact for purchasing palm oil, the crop (denoted by c in the equations), from the region g, which here is Aceh, Indonesia. Where, HAc,i represents the harvest area of palm oil per pixel, i, in hectares; Pc,i is the production of palm oil in each pixel in tonnes; and Sc,g is the amount of palm oil sourced from the Aceh region in tonnes.

Landscape-level impacts

Landscape-level impact indicators (e.g. Figure 3) are calculated in the following way:

• For unknown, country, administrative region or aggregation point location types, the production map is buffered using a radius kernel prior to using it as the weighting layer, in order to capture areas nearby to producing regions.
• The impact factor is calculated as the production weighted average within the sourcing geometry using the buffered production map. If there is no overlap between the production map and the sourcing geometry, the impact factor is calculated as the area average for the sourcing geometry.
• To avoid double counting proximal landscape-level impacts, the impact is distributed across all the crops produced in the sourcing geometry, and impacts are allocated in proportion to each crop’s area footprint.

The choice of buffer size by which to analyze impacts in the proximity of production systems for landscape level indicators defaults to 50km in version 0.1 of LandGriffon. Future developments may implement a more complex approach to distances, for example, varying with the material, the characteristics of the production landscape, and/or the administrative region.

Figure 3. Example of a landscape-level impact calculation of the weighted deforestation risk calculation for purchasing 1 ton of Palm oil sourced from Aceh, Indonesia.

What impacts are already included?

In LandGriffon, farm- and landscape-level impact calculations are implemented in a modular way so that it is easy to integrate new indicators with supply chain ingestion and spatial geolocation framework. Version 0.1 of LandGriffon includes a baseline set of impact indicators (Table 1). Further descriptions of indicators objectives, calculations, limitations, and next steps are provided in Annex 1 of the full methodology downloadable from our website.

These indicators are currently focused on production impacts, as opposed to lifecycle impacts. So, although LandGriffon has taken some inspiration from Life Cycle Assessment (LCA), there will be implicit impacts associated that are not captured by production-focused indicators.

To learn more about the science and technology of LandGriffon, you can download the full methodology from our website.

Interested in what LandGriffon could do for your company?

Contact us now at hello@landgriffon.com

LandGriffon is developed by Satelligence and Vizzuality, and advised by the Stockholm Environment Institute and their Trase Initiative.

Thank you Mike Harfoot, Elena Palao, Francis Gassert and Rens Masselink for preparing the methodology.