UAV thermal imagery contribution to high throughput field phenotyping of apple tree hybrid population and characterization of genotypic response to water stress
Résumé
High-throughput field phenotyping can be performed by using high resolution multispectral imagery. This presentation focuses on Unmanned Aerial Vehicle (UAV) sensed thermal imagery used to assess the response of an apple population to drought and analyze the genotypic variability of stomatal behavior. Studies were performed in an experimental apple orchard located in Southern France submitted to different water regimes. For remote image acquisition, UAV flights were performed at different dates, hours and altitudes with a thermal camera on board (0.10-0.27m spatial resolution). Temperature of different reference ground targets (hot, cold, wet and dry bare soil) was continuously measured by thermo-radiometers for image radiometric calibration. To assess the effect of image resolution and that of vegetation cover fraction, a sample of 18 apple trees was chosen and the mean canopy temperature (Ts) in tree central zone and its variability were measured. As distortions were revealed in Ts, it seemed advisable to separate mixed pixels (including shaded leaves and soil) from well-illuminated vegetation. Considering the effect of altitude, standard deviation of Ts increased according to the image resolution, and this was particularly true where resolution was close to leaf average size (0.10m2). Thanks to the ultra-high resolution of remote images obtained, and beyond capacity of the approach to delineate efficiently each individual tree within the whole trial, it was possible to analyze inter- and intra-canopy thermal variations. By using image-based vegetation and stress indices as phenotypic variables, genetic dissection of the traits is currently undertaken and first results will be presented.