Résumé
Light distribution within fruit trees has a major influence on the distribution of fruit quality and diseases. A method for estimating light partitioning at the intra-tree canopy scale is presented. It is based on the three-dimensional (3D) reconstruction of the tree canopy at the organ scale. First, the trees are 3D-digitised with a magnetic device at current-year shoot scale. Second, shoot foliage is 3D-reconstructed by using allometric relationships and leaf angle distributions built from sampled shoot digitising at leaf scale. Third, false colours are attributed to plant components, e.g., shoot types, in order to distinguish them on the virtual plant images. Fourth, orchard scenes are created with a coloured target tree at the centre and black neighbour trees shading the target tree. Fifth, images of the virtual scene are created with view directions set to those of the light sources, namely according to the sun course for direct radiation and a set of sampled directions over the sky hemisphere for diffuse radiation. Sixth, light interception by each coloured plant component is computed from processing the scene images, namely coloured pixels counts. The method was used with eighteen 13-year-old Golden Delicious apple trees trained in vertical axis, ycare and drilling systems. Light interception was computed at fruit scale (using one colour per fruit). As an indirect validation of the method, simulated fruit irradiance was related to actual fruit sugar content. At the individual fruit scale, r² coefficients were about 0.3–0.4. After grouping fruits in irradiance classes, r² coefficients were mostly above 0.85. The method was therefore found suitable for estimating light distribution within fruit trees.