Skip to content

Soy

View or edit on GitHub

This page is synchronized from trase/models/brazil/soy/README.md. Last modified on 2025-12-13 00:30 CET by Trase Admin. Please view or edit the original file there; changes should be reflected here after a midnight build (CET time), or manually triggering it with a GitHub action (link).

Brazil Soy

Methods documentation: Brazilian crop commodities review (Google Docs)

Some example model running times (load + run, but not preparation):

Year Runtime Peak Memory
2019 50 minutes
2020 50 minutes
2021 4 minutes 3.8 GB
2022 4 minutes 4.1 GB

Model Changelog

This section documents any changes which would be of interest to a consumer of the model output. We do not document backend changes (technical debt, refactors, etc.) which have no impact on the data itself.

2.6.1:

November 2024:

  • We have made some improvements to the trader hierarchy. In particular, these result in improvements to the Zero Deforestation commitments in 2018
  • RIBEIRAO SA is now part of the RISA trader group (with "UNKNOWN" deforestation commitments)
  • COOPERATIVA AGROIND DOS PROD R DO SUDOESTE GOIANO is now a label of the COOPERATIVA AGROINDUSTRIAL DOS PRODUTORES RURAIS DO SUDOESTE [TRADER | BR-TRADER-02077618 | BRAZIL] trader (with "NONE" deforestation commitments)
  • Used MapBiomas collection 9 for deforestation metrics (was previously collection 6)
  • We have altered the way that embedding is done on UNKNOWN flows

July 2024:

  • Added exporter group to Synacomex data, since the matching in 2019 and 2020 was resulting in many more unknown exporter groups than we saw in the v2.6.0 run.
  • Fixed an issue with reading of special_cases.csv where it was not including entries where year = "all".
  • Added new special case for Bunge Alimentos' headquarters (84046101000193).
  • Added exporter group to 2019 and 2020 run, to allow for new-style metric embedding.
  • Reclassify the CNAE 0163-6/00 ("post harvest activities") to level 5 for 2021 and 2022. Previously it was level 1.

Before July 2024:

  • Introduced SICARM into the decision tree.
  • Fix a potential bug where duplicate exporter CNPJs were perhaps introduced - unsure about this though (01ebd).
  • Note the following differencs to the 2021 years:
  • Trader names are not imputed using port/exporter stickiness as they are in 2019 and 2020.
  • Synacomex vessels are also not matched for 2021 and 2022.
  • Does not use the SECEX Cadastro dataset since it is not longer available. See check_secex.py for a script analysing the impact and the data catalogue entry for brazil/auxiliary/secex.
  • Uses new CNPJ data processed from Receita Federal do Brasil (RFB), fetched in July 2023. See the entry on trase-storage/brazil/logistics/cnpj/receita_federal_do_brasil in the data catalogue.
  • Use 2023 SICASQ data (was previously 2017) for both 2021 and 2022.
  • In 2020 Abiove's crushing capacity is based on JJ Hinrichsen's report in 2021 (d25925).