Team Europe Iniciativa da UE para Cadeias de Valor Livres de Desmatamento

This document provides methodological guidance for the collection of high-quality reference data through visual interpretation of very high resolution (VHR) imagery to support supervised classification of remotely sensed data for thematic crop type and crop probability mapping. It defines a consistent framework for visual interpretation, digitisation, and quality control of planted areas (PAs) using VHR satellite imagery, with a focus on perennial crops in heterogeneous agricultural landscapes.
The guidelines are designed to support applications that require retrospective labelling for a fixed reference year, as is the case for regulatory frameworks such as the EU Regulation on deforestation-free products (EUDR), where direct field verification is no longer possible. In this context, reference-year VHR imagery is complemented by multi-year VHR imagery to support crop identification, confirm crop persistence, and clarify PA boundaries where single-date imagery is insufficient. A sampling-unit-based, wall-to-wall digitisation approach is adopted to optimise data collection for developing segmentation models, preserving spatial context and enabling scalable generation of diverse training samples.
While the guidelines are formulated in generic terms and are applicable across perennial crops and regions, they are illustrated using coffee mapping in Kenya as an implementation example. The work was developed through collaboration between European and Kenyan institutions within the framework of the European Commission’s Teams Europe Initiative and the Technical Facility on deforestation-free supply chains, supporting Kenya’s preparedness for implementation of the EUDR.