Detection of crop water stress using UAVs: practices and new insights in utilizing uncooled thermal sensors

Farmers increasingly use drones to monitor the health of their crops. One of the most valuable things a drone can measure is how much water a crop has: a plant running short of water heats up, much like a person does. Special thermal cameras can "see" this heat and reveal water stress before it becomes visible to the eye, helping growers irrigate more precisely and waste less water.
The catch is that the lightweight thermal cameras carried by drones are affordable but not very accurate: their readings drift as the camera warms up and shift with wind, temperature and flight height. This thesis uncovers why these errors occur and develops practical methods, from laboratory tests to in-field calibration, to correct them. Combining the corrected thermal data with colour and near-infrared imagery, it shows how reliably mapping crop water stress can also improve estimates of crop growth, making low-cost thermal drones a trustworthy tool for precision agriculture.