Water Stress Detection Using Thermal and Hyperspectral Imagery Acquired from Unmanned Aerial Vehicles.

This work presents an overview of the current progress in water stress detection using thermal and hyperspectral cameras on board Unmanned Aerial Vehicles (UAV). Recent studies have demonstrated that micro-hyperspectral and thermal cameras carried by lightweight remotely piloted aircraft (RPAS) enable the generation of several narrow‑band spectral indices related to crop physiological condition, canopy temperature, as well as the quantification of the chlorophyll fluorescence emission linked to plant photosynthesis. Work conducted with two different fixed-wing platforms for this purpose will be presented, ranging between 2 and 5-m wingspan to fly autonomously at altitudes between 500 and 1500 ft with endurance time to cover up to 1500 ha per flight. The miniaturized thermal cameras used in the experiments yield pixel resolutions between 20 and 50 cm depending on the flight altitude, and are calibrated in laboratory and using targets in the field. The micro‑hyperspectral imager flown acquires between 260-360 spectral bands and 3 to 6 nm FWHM imagery depending on the acquisition mode used, obtaining 40 cm spatial resolution in the 400‑1000 nm spectral range. Radiometric calibration, atmospheric correction and imagery ortho-rectification methods are conducted using commercial solutions and software developed at QuantaLab – IAS – CSIC laboratory. The results obtained using these miniaturized cameras demonstrate that sun-induced fluorescence, canopy temperature and narrow-band indices show good relationships with physiological indicators such as stomatal conductance, stem water potential (SWP) and leaf photosynthesis. The indices calculated are related to xanthophyll, carotenoids, anthocyanins, and chlorophyll a+b absorption using optical indices such as PRI, TCARI, red edge, as well as with new narrow-band indices. Methodologies to calculate the crop water stress index (CWSI) at tree crown level in orchards and the relationships obtained between thermal and hyperspectral indicators with ground-measured SWP, stomatal conductance and assimilation rates will be discussed. For operational purposes, automatic object-based image analysis methods applied to the thermal and hyperspectral imagery will be described, enabling the generation of thematic maps of water (CWSI) and nutrient stress in orchards and vineyards at individual crown level without soil and background effects in less than 24 hours.