1. National time series (2008-2024)
National and biome-level deforestation totals across datasets track differences over time (2008-2024). MapBiomas shows more deforestation across all years up to 2021. The difference relative to GLAD is consistently negative from 2021. In 2024, deforestation reached the lowest value in MapBiomas.
2. Biome-level differences
All biomes show lower deforestation in MapBiomas in 2024, with the Pampa and Atlantic Forest showing the most striking differences.
3. State-level differences
In 2024, RS shows a large difference between MapBiomas and GLAD. MapBiomas reported less than 200,000 ha of soy deforestation compared with GLAD in RS.
4. Municipality-level map
Differences (proportional [%] difference: GLAD - Mapbiomas) in soy deforestation between datasets at municipal level, with a toggle for absolute deforestation values.
5. Municipality image comparison (2024)
In 2024, deforestation is clearly higher in GLAD, particularly in the regions highlighted below. The main reason for lower deforestation in MapBiomas is the difference in mapped soy area. For example, in Sao Joao del Rei (MG) there is no planted soy. This strongly affects the 2024 results and pushes MapBiomas deforestation down. Slide to compare MapBiomas (left) and GLAD (right) deforestation maps for selected municipalities.
Notes
- We tested different Minimum Mapping Unit (MMU) values ranging from 222 pixels (20 ha as per Song et al. (2021)) to 0 pixels, but soy area is still underestimated in 2024 even when no MMU is applied.
- We also tested different MMU thresholds for territorial deforestation in the MapBiomas deforestation and secondary vegetation product, ranging from 70 pixels (6.2 ha as per the PRODES MMU) to 0 pixels. Reducing the MMU did not increase deforestation in 2024.
- Given the clear differences in the most recent year, and the risk of overlooking relevant producing municipalities, it would be more prudent to advise CGF companies to use the GLAD soy area combined with PRODES deforestation (2024).