UAV-based infrared thermography for laminar–turbulent transition detection on wind turbines in operation: quantifying motion-blur effects using blade image velocity

Abstract. This study investigates the applicability of uncrewed  aerial vehicle (UAV)-based infrared thermography (IRT) for aerodynamic flow visualization on operating wind turbines, with a focus on the localization of the laminar--turbulent transition. While UAV deployment enables flexible, non-contact measurements at large stand-off distances, the use of lightweight microbolometer cameras introduces limitations related to temporal response and motion blur induced by high blade image velocity. A Gaussian error-function-based approach is employed to localize blade edges and transition features in thermographic images. Controlled laboratory experiments are conducted to isolate the influence of motion blur over a wide range of blade image velocities. The results show that increasing blade image velocity leads to a progressive broadening of temperature gradients and a corresponding increase in localization uncertainty. At high image velocities, the underlying intensity profiles deviate from the assumed model shape, resulting in a marked loss of robustness in the edge-detection procedure. To mitigate these effects, image deblurring based on Wiener deconvolution is applied using a point-spread function derived from the exponential response of the microbolometer detector. The deblurring approach significantly improves the stability of the evaluation and reduces the transition-location uncertainty by approximately a factor of five at high blade image velocities. The methodology is further applied to field measurements on a 1.5 MW wind turbine. The results demonstrate that transition-related thermal signatures can be detected under operational conditions and that deblurring substantially enhances the visibility of flow features, particularly in regions of high blade image velocity. Field-based uncertainty estimates further show that, at high blade image velocities, deviations from the assumed signal model become the dominant source of error, while deblurring primarily improves the robustness of the transition localization rather than uniformly reducing uncertainty. Thus, the findings identify motion blur as the dominant limitation for quantitative UAV-based IRT measurements and demonstrate that its impact can be effectively reduced by appropriate post-processing. The presented approach provides a framework for estimating motion-blur-induced uncertainty and defines practical limits for transition localization in airborne thermographic measurements.