Unbiasing entomological kHz Scheimpflug LIDAR data

Abstract: Insects can be mapped using a method called Scheimpflug Lidar (LIght Detection And Ranging). Sending out light in the near infrared range and detecting the backscattered echo, it is possible to resolve flying organisms in range and time. Insect observations can be distinguished from the background signal through adaptive thresholding. The detected signal will contain two components; one from light that has scattered on the insect body, and the second due to the wing-beats cycles of the insect. Further, the shape of the wing-beat signal may contain several harmonics due to non harmonic motion and specular reflections. The wing-beat frequency (WBF) of an insect can be found using a parametrization model where the wing-beat component of signal is projected on a number of harmonic functions. Projections using different possible WBF are done, and the frequency that generates the lowest residual is chosen as the WBF of the signal. Previously, the model has been used in short frequency ranges around a WBF estimated from the signal and the power spectrum of the signal since the model is biased towards minimum possible frequency and Nyqvist frequency. In this thesis, three different compensation methods to overcome the frequency preferences of the model has been investigated with promising results. Further, range biasing and detection limits of the system has been investigated, showing that a longer exposure time makes it possible to detect signals with almost five times as low intensity and reduced the apparent size of the insect. Regarding the range biasing, the system generally detects more observations at short ranges due to higher sensitivity. However, when the range increases a higher amount of observations with long transit time is detected.