This is what a real screening looks like.

Six coffee fields near Sarutaia, Sao Paulo, declared together as one plot of 386ha, which FAQ 1.14 allows for parcels belonging to one farm. Zoom in far enough and the planting rows are visible. The checkbox overlays the EU's 2020 forest map, the reference layer the screening tests against.

A screening checks six satellite loss datasets through WHISP (Open Foris, FAO), the EU's 2020 forest map, and national land-use data. The 58 ha coffee field next to this block came back deforestation-free:

Stored verdict: Cleared · screened 2026-08-15 · WHISP 3.0.0a16

No forest at the 31 Dec 2020 cutoff: the EU Forest Observatory 2020 map shows 0 ha of forest on this plot and none of the JRC 2020 forest types (primary, naturally regenerating or planted, Art. 2(8)-(11)) is present. Confirm the land was already in agricultural use before the cutoff and record your determination.

The EU map shows no forest on this field at the cutoff, and no dataset reports loss after it. Most plots in a portfolio come back like this and need no further work.

Stored verdict: Needs review · screened 2026-08-15 · WHISP 3.0.0a16

Cannot conclude deforestation-free yet: a loss dataset reports up to 0.50 ha of post-cutoff tree-cover loss intersecting the polygon (tree-cover loss is not automatically deforestation, so review it against the parcel boundary and pre-2020 land use before concluding); the plot was forest at the 31 Dec 2020 cutoff and shows a post-cutoff conversion signal, so confirm it was already in agricultural use before the cutoff.

Two readings decide this. 30.6 ha of tree-cover loss was detected in 2020, before the cutoff, so it carries no obligation. 0.50 ha was detected across 2021 and 2022, after it:

The reviewer gets context with the signal: MapBiomas classifies 234.3 ha of this block as coffee in 2020, and INPE's PRODES shows no deforestation here after the cutoff. The block is not failed on 0.50 ha, and it is not cleared automatically either. The EUDR determination is made by the operator, and the screening stores the evidence it will rest on.

For a held plot the product also reads monthly Sentinel-2 greenness back to 2017 and compares it with the plot's own pattern from before the cutoff. On this block the two findings point in different directions, and the view states both: twelve months since the cutoff read lower than the pre-cutoff minimum, and cloud cover does not explain them; at the same time, the fitted level across the cutoff changes by 0.04, which is within the uncertainty of the fit. The reviewer takes that to the imagery.

greenness through the cutoff