Thermal Drone Images as a Predictor of Soil Moisture Values

Drones are one of the latest additions to the arsenal of technologies for precision farming. Their use is still restricted but is expected to grow as farmers get used to them, and they get positive peer reviews of their usefulness and more practical use cases for them. To provide a practical use case, we have studied the use of thermal drone images as a predictor of soil moisture values. We used a SenseFly eBee X fixed wing drone to monitor a barley field in Saarijärvi, Central Finland, during the growing season of 2022. Images were processed with Pix4D software and analyzed using Python programming language. In addition to thermal images, we used Soil Scout soil moisture sensors to acquire actual soil moisture values. The images were compared with the soil sensor data to see how well and under what conditions they predicted actual soil moisture values. The results showed that without any preconditions plain thermal images do not predict soil moisture values well, but with certain conditions mild predictability can be achieved.