Thread by @erasmuszu

In non-COVID news, new science!

We show that a lot of information is hidden in the commonly used ‘representative’ environmental footprints of products. For accurate assessment you need supply chain specific data & @TraseEarth data allows you do this at large scale. 1/

Calculating the environmental/carbon footprint of products is hard, not least because many products are internationally-traded, which means that you may need data on production practices from the other side of the world.

Normally, people estimate the impacts embedded in trade in one of a few ways – using some kind of bilateral trade data (e.g. Environmentally-Extended Multi Regional Input Output, or EE-MRIO, models) or using product-specific life cycle assessments (LCA).

The 1st method gives you a global picture but oft relies on national-level data, which mask a lot. Brazil is twice the size of the EU, so if estimating the impact of, say, Brazilian soy in animal feed, you risk lumping a lot of different production systems together

The second approach (LCA) is very detailed, but it also requires a mountain of data – which is time consuming and costly to collect, and insight may be difficult to generalize to other products.

So, we used @TraseEarth’s subnational data to bridge this gap, calculating the carbon footprint of Brazilian soy exports for 2010-2015 based on ~90,000 individual flows of beans, oil & meal tracked from the municipality of origin through to specific international markets

We find huge variability in emissions across sourcing areas, supply chain stages & countries of import. There is no ‘average’ soy & we shouldn’t treat it as such: when companies avoid Brazilian soy, they oversimplify & punish sustainable producers as well as the bad

Here you can see how exports from MATOPIBA states and Pará are dominated by land use change emissions, which is completely different from, say, Goiás, where emissions are lower and mostly driven by domestic transport.

We also split this per importing country or region – here you can see how half of the emissions imported to the EU from Brazilian soy are because of land use change, which makes up only a quarter of emissions imported by China.

This resolution data is, (I think!) a big improvement over previous work – below you can see how previous typically studies provide point estimates of emissions/ton product, where we can now disaggregate that for each specific supply chain, giving a range of estimates

Two notes: [i] we showcase Brazilian soy but the approach applies to any commodity with high resolution trade data. @TraseEarth is constantly expanding to new products. [ii]
We split the data per country but can also do it per company, a win for corporate emissions accounting.

Open-access article: sciencedirect.com/science/articl…

Press release (EN): uni-bonn.de/news/114-2020
Press release (DE): uni-bonn.de/neues/114-2020