Case Study: CERFO (Centre d’enseignement et de Recherche en Foresterie de Sainte-Foy) achieves automated tree-level species classification for targeted forest management with Trimble eCognition and Pro 6H

eCognition helps researchers get to the root of classifying individual trees

The Centre D’enseignement et de Recherche en Foresterie de Sainte-Foy (CERFO) in Quebec — a region where nearly half the province is forested — needed a faster, more accurate way to classify individual trees in dense, heterogeneous forests. Traditional methods rely on manual delineation at the stand level, which is time-consuming and error-prone; CERFO set out to automate species-level mapping for Broadleaf and Conifer trees across challenging terrain in the Kenauk Nature Preserve.

Using Trimble eCognition with WorldView-3 imagery, a lidar-derived canopy height model, and field reference data collected with Trimble Pro 6H GPS receivers (515 trees surveyed, 338 samples), the team segmented individual crowns (>17 m) and classified 11 tree types in about two hours. The automated workflow delivered high precision (94% overall accuracy for Conifers), produced species-specific GIS layers for targeted silviculture and harvesting, and prompted further refinement and adoption by forest managers.

CERFO

Mathieu Varin

Remote Sensing Laboratory Manager