Key finding
The study evaluated 21 global forest and tree cover datasets against three practical criteria: spatial resolution adequate for farm-level assessment, temporal coverage including the EUDR baseline date of 31 December 2020, and consistency with the EUDR's forest definition (minimum 0.5 hectares, at least 10% canopy cover, trees at least 5 meters tall). Only eight datasets met the basic spatial, temporal, and definitional criteria. Only two, derived from Sentinel-2 and Planet high-resolution imagery, fully matched all elements of the EUDR forest definition.
The study found significant regional inconsistencies. Some datasets performed well in humid tropics but poorly in dry forests and savannahs. Others systematically overestimated forest cover by including plantations as natural forest, or underestimated by missing dry forests entirely. Reliance on a single forest map creates significant compliance risk: different maps give different answers for the same location.
What the research asks
The European Union's Deforestation Regulation (EUDR) requires companies to prove their products were not grown on land deforested after 31 December 2020, placing unprecedented reliance on global forest maps for regulatory compliance. But global forest maps were not designed for this purpose, and different maps define forests differently, use different resolutions, and cover different time periods. This study asked: of the available global forest and tree cover datasets, how many are actually suitable as reference tools for EUDR-compliant deforestation monitoring, and what risks arise from relying on a single map?
What it finds
The assessment applied three practical criteria to each dataset: spatial resolution adequate for farm-level assessment, temporal coverage that includes the EUDR baseline date of 31 December 2020, and consistency with the EUDR's forest definition (minimum 0.5 hectares, at least 10% canopy cover, trees at least 5 meters tall). The results are sobering.
Only eight of the 21 datasets met the basic spatial, temporal, and definitional criteria for EUDR compliance checks. And only two datasets, the high-resolution forest maps derived from Sentinel-2 and Planet imagery, fully matched all elements of the EUDR forest definition. The rest either lacked the spatial resolution to detect small-scale deforestation (below 0.5 hectares, a common scale for smallholder farming), used different forest definitions (for example, defining forest as any land with more than 30% tree cover, which would miss significant deforestation), or lacked a reliable baseline for the 2020 cutoff date.
The study also found significant regional inconsistencies. Some datasets performed well in the humid tropics but poorly in dry forests and savannah ecosystems. Others showed systematic overestimation of forest cover in some regions (including plantations and agroforestry as natural forest) and underestimation in others (missing dry forests entirely). The risk of false positives (flagging deforestation where none occurred) and false negatives (missing genuine deforestation) varied widely by dataset and region.
Crucially, the authors found that reliance on a single forest map creates significant compliance risk. Different maps gave different answers for the same location, meaning a company could pass a due diligence check using one dataset and fail using another, with no clear way to know which is correct.
Why it matters
This paper is essential reading for every sustainability professional preparing for EUDR compliance. It provides a practical evidence base for the common-sense recommendation that companies should not rely on any single forest map, but should combine multiple datasets, integrate national and regional reference information, and maintain a clear audit trail of which maps were used and why.
The paper also highlights a deeper structural challenge: the EUDR's forest definition, borrowed from FAO's statistical framework, was not designed for satellite-based compliance verification at the scale of individual farm plots. The mismatch between policy definitions and monitoring capability creates real operational risk for companies and real enforcement challenges for regulators. The authors' call for harmonization of forest definitions across monitoring frameworks speaks directly to the work of Earth's community, who are building the monitoring infrastructure for a deforestation-free economy.