Thermal sensors mounted on unmanned aerial vehicles (UAVs) have been used for crop evaluation. The assembly of multiple UAV images into a composite orthomosaic image that represents an agricultural field is common. The creation of these composite orthomosaic images is predicated on the assumption that the variable of interest, for example, plant height, does not change in the interval between the first image and the last image. Canopy temperature is continuously variable. Thus, in a composite thermal image, the temperature differences between any two points can include both the temperature differences of the two locations, as well as the temperature change occurring during the time the images were captured. A modeling approach was used to investigate these temporal thermal artifacts in cotton (Gossypium hirsutum L.). The objectives were to analyze the average amount of thermal variation that can occur across a field, identify times of maximum variation, and examine the magnitude of these thermal artifacts across the temporal scale of a typical drone mission. As the time between the first image and last image increased from 15 minutes to an hour, the temporal thermal distortion increased from 0.7 to 1.4 °C for a high-water treatment and from 0.9 to 2.8 °C for a low-water treatment. The simulated flights from 1030 to 1230 h had the largest variation and the 1230 to 1430 h and 1430 to 1630 h flights had less thermal variation. Depending on the application, the thermal distortion could range from negligible to potentially significant.
