Challenges and Best Practices for Deriving Temperature Data from an Uncalibrated UAV Thermal Infrared Camera

Abstract

Miniaturized thermal infrared (TIR) cameras that measure surface temperature are increasingly available for use with unmanned aerial vehicles (UAVs). However, deriving accurate temperature data from these cameras is non-trivialsince they are highly sensitive to changes in their internal temperature and low-cost models are often not radiometrically calibrated. We present the results of laboratory and field experiments that tested the extent of the temperature-dependency of a non-radiometric FLIR Vue Pro 640. We found that a simple empirical line calibration using at least three ground calibration points was sufficient to convert camera digital numbers to temperature values for images captured during UAV flight. Although the camera performed well under stable laboratory conditions (accuracy ±0.5 °C), the accuracy declined to ±5 °C under the changing ambient conditions experienced during UAV flight. The poor performance resulted from the non-linear relationship between camera output and sensor temperature, which was affected by wind and temperature-drift during flight. The camera’s automated non-uniformity correction (NUC) could not sufficiently correct for these effects. Prominent vignetting was also visible in images captured under both stable and changing ambient conditions. The inconsistencies in camera output over time and across the sensor will affect camera applications based on relative temperature differences as well as user-generated radiometric calibration. Based on our findings, we present a set of best practices for UAV TIR camera sampling to minimize the impacts of the temperature dependency of these systems.

A Review of Current and Potential Applications of Remote Sensing to Study the Water Status of Horticultural Crops

“…Another challenge with the use of thermal cameras is the temporal drift of the DN values within successive thermal images, especially with uncooled thermal cameras [143]. Due to the lack of an internal cooling mechanism for the microbolometer detectors, DN values registered by the microbolometers experience temporal drift i.e., the registered DN values for the same temperature target will drift temporally.…”

“…Thus, the thermal image can be unreliable especially when the internal temperature of the camera is changing rapidly, such as during camera warmup period or during the flight when a gust of cool wind results in cooling of the camera. To overcome this challenge, the user may need to provide sufficient startup time before operation (preferably 30-60 min) [102,[143][144][145], shield the camera to minimize the change in the internal temperature of the camera [142], calibrate the camera [146][147][148][149][150][151][152][153], and perform frequent flat-field corrections.…”

“…Furthermore, for low measurement errors (under 1 • C), thermal data requires consideration to the atmospheric transmittance [18,102]. Flying over a few temperature reference targets placed on the ground reduces the temporal drift of the camera [142,143,181]. Temperature accuracy within a few degrees was achieved by flying over the targets three times (at the start, middle and end of UAS operation) and using three separate calibration equations for each overpass [142].…”

“…The manufacturer stated accuracies (generally ±5 • C) can be sufficient to access the field variability and to detect "hotspots" of water status. However, the aforementioned calibration and correction of the thermal cameras are required for quantitative measurement as a goal [143]. In this regard, current challenges and best practices for the operation of thermal cameras onboard a UAS is provided in the literature [143].…”

UAV-based Multispectral & Thermal dataset for exploring the diurnal variability, radiometric & geometric accuracy for precision agriculture

“…Multiple studies refer to the capabilities of these sensors (Elarab et al 2015;Gago et al 2015;Haghighattalab et al 2016;Holman et al 2016;Potgieter et al 2017;Raeva, Šedina, and Dlesk 2018;Sankaran et al 2015;Shi et al 2016;Zhang and Kovacs 2012). A few studies refer to the limitations and best practices, to achieve high-quality data (Aicardi et al 2016;Assmann et al 2018;Kelly et al 2019;Maes, Huete, and Steppe 2017;Peña et al 2015;Torres-Sánchez et al 2013). However, in existing studies the emphasis is on the differences or complementary use of different sensors while the effect of the diurnal differences in the data is not fully explored.…”