Résumé
As breeding fruit varieties for adaptation to drought becomes a new challenge, it is necessary to assess genetic resources and new cultivars under water restriction. Availability of affordable high-resolution sensors acquiring images in multiple narrow visible (VIS) and near infrared (NIR) wavebands allows accurate computation of vegetation and/or stress indices (e.g., NDVI, GNDVI, MCARI2, PRI). These indices reveal phenotypic modifications of individual plant cover structure and/or leaf functions. Moreover, as transpiration is limited by stomatal closure, increase in canopy surface temperature (Ts), measured by thermal infrared (TIR) camera, is a real-time indicator of plant response to water deficit, and TIR imagery was recently used for calculation of proxies for stomatal closure in apple tree (Ts_excess = Ts-Tair). In the experiments presented, VIS/NIR and TIR signatures were acquired on apple trees submitted to normal and limited irrigation scenarios during summer, while soil water status was continuously and spatially monitored. Multispectral and thermal sensors were installed on an unmanned aerial systems (UAS) and aerial images were acquired on a weekly base, allowing the establishment of temporal variation of indices, which is of particular interest since limitation or cessation of irrigation causes non-concomitant progression of water stress across cultivars. A semi-automated chain of image processing was developed, to compute indices within individual trees. Parallel to airborne imagery, leaf and stem water potential were measured in planta. Relationships between vegetation indices and field measurements were established, highlighting particular relevance for the information acquired by a 6-channel VIS/NIR narrowband sensor. The phenotyping procedure proposed paves the way to further studies in apple genotypic collections, while typology of apple cultivar response to soil water deficit was shown in assessment assays.