This manuscript presents a comprehensive methodology to obtain Thermal, Visible and Near Infrared ortho-mosaics, as a previous step for the further image-based assessment of response to water stress of an experimental apple tree orchard. Using this methodology, multi-temporal ortho-mosaics of the field plot were created and accuracy of ortho-rectification and geo-location computed. Unmanned aerial vehicle (UAV) flights were performed on an irrigated apple tree orchard located in Southern France. The 6400 m² plot was composed of 520 apple trees which were disposed in 10 rows. In this field set-up, five well irrigated rows alternated with five rows submitted to progressive summer water constraints. For remote image acquisition, on 4th July, 19th July, 1st August and 6th September UAV flights with three cameras onboard (thermal, visible and near infrared) were performed at solar noon. On 1st August, five successive UAV flights were carried out at 8, 10, 12, 14 and 16 h (solar time). By using selfdeveloped software, frames were automatically extracted from the recorded thermal video and turned in the right image format. The temperature of four different targets (hot, cold, wet and dry bare soil) was continuously measured by the IR120 thermoradiometers during each flight, for radiometric calibration purpose. Based each on thirty images, all ortho-mosaics were successfully obtained. As high spatial resolution imagery requires high precision geo-location, and the root mean squared error (RMSE) of each ortho-mosaic positioning was calculated in order to assess its spatial accuracy. RMSE values were less than twice the pixel size in every case, which allowed a precise overlapping of the mosaics created. Canopy temperature data extracted from thermal images for showed significantly higher temperatures in water stressed trees compared to well irrigated, difference being related to severity of water stress. Thanks to the ultrahigh resolution of remote images obtained (<0.1m spatial resolution for thermal infrared images), and beyond its capacity to delineate efficiently each individual tree, the methodology presented here will also make it possible the analysis of intra-canopy variations and the accurate calculation of vegetation and water stress indices.