Surface temperature variations observed from a thermal infrared camera mounted on a hovering UAV platform

Abstract

<p>Land surface temperature (LST) is crucial information that helps to understand and assess the interactions between the surface and the atmosphere. LST is a key parameter used in various applications including studies of irrigation, water use, vegetation health, urban heat island effects, and building insulation. In addition to several satellites that provide periodic images of surface temperature, unmanned aerial vehicle (UAV) platforms have been adapted to obtain higher spatio-temporal resolution thermal infrared (TIR) data. In fact, numerous research studies have investigated the accuracy and the processing method of UAV-based TIR images given its complexity and sensitivity to ambient conditions. However, the surface temperature is characterized by continuous and rapid variation over time, which is difficult to take into consideration in the processing of UAV-based orthomosaics. Here, we quantify this variation and discuss the environmental factors that lead to its amplification. Thermal images were collected over a fixed hovering position during periods of 15-20 min, representing the common duration of UAV flights. At different times of the day, we flew at different altitudes over sand, water, grass and olive trees. Before the quantification of the surface temperature variation, the thermal infrared data were evaluated against field-based measurements using calibrated Apogee sensors. The evaluation showed a significant error in the UAV-based thermal infrared data linked to wind speed, which increased the bias from -1.02 to 3.86 °C for 0.8 to 8.5 m/s winds, respectively. The assessment of the LST values collected over the different surfaces showed a temperature variation while hovering ranging between 1.4 and 5 °C. In addition to wind effects, temperature variations while hovering were strongly linked to solar radiation, specifically radiation fluctuations occurring after sunrise and before sunset. This research provides insights into the LST variation expected for standard UAV flights of 15-20 min under different environmental conditions, which should be taken into account during UAV-based thermal infrared data processing and may help interpret and quantify inconsistencies in UAV-based orthomosaics of LST.</p>