Coherent wind-lidar for optical turbulence measurement

Time-accurate information on the turbulent parameters of the atmospheric boundary layer is relevant for a variety of technical applications, i.e., the turbulent kinetic energy dissipation rate (TDR) or the turbulent index of refraction. For instance, these include the real-time detection of clear-air turbulence for turbulence avoidance or the atmospheric propagation of laser beams. In the context of this work, a lidar system was built to provide this data in near real-time based on preprocessing hardware developed prior to this thesis. Further, the necessary control software and optimized evaluation algorithms were developed. An essential part was the completion of an optical system for the transmission and reception of laser pulses and backscattered light, respectively, in the near infrared spectrum. Mono-, as well as bistatic designs were tested and alignment processes developed. The robustness and operability of the monostatic design proved to be superior. A coherent detection method was used due to the significant downshift in bandwidth that can be achieved. Field tests were performed to demonstrate the capabilities of the lidar system. Although only static measurements were conducted, that is the wind speed in the direction of the line-of-sight was recorded, interesting measurement data was obtained. A TDR-estimator based on the analysis of the velocity density spectrum was used to determine height-resolved values for the turbulent kinetic energy dissipation rate. From this, values for the turbulent index of refraction could be derived. An improved and automated algorithm will be presented for the estimation of the TDR, which can operate on velocity data even from short time intervals. For this purpose, a method was developed that can ignore the influence of statistically dependent effects and achieve a minimization of the relative error for the TDR. Another novel approach, the use of a convolutional neural network for the velocity estimation from lidar spectrum data, will be investigated. In this context, the distributions of the parameters for a synthetic data generation model are examined in more detail. Selected field campaigns will be presented and optical turbulence parameters compared to the Hufnagel-Valley model. Qualitatively, it will be shown on this real-world data that the TDR-estimator indeed finds optimized solutions in the velocity density spectrum