Turbulence measurements with fixed and floating profiling wind lidar

Knowledge about turbulence in the wind helps estimate and optimize the profitability of wind turbines. Light detection and ranging (lidar) is a cost-efficient, flexible, and accurate remote sensing technology for measuring wind velocities. Comparisons of mean wind data from lidar and in situ anemometry show good agreement, but estimates of turbulence such as turbulence velocity spectra and turbulence intensity deviate significantly. In this thesis, we present methods to predict lidar-derived turbulence velocity spectra. For the case of a velocity–azimuth display (VAD) scanning continuous-wave wind lidar, we introduce a numerical model that filters a spectral tensor so the resulting spectra resemble those derived from lidar measurement data. For a Doppler beam swinging (DBS) pulsed wind lidar, we sample computer-generated turbulence data in a similar way to how a lidar measures real wind velocities. Averaging the results from many data series leads to comparable results between simulated and measured lidar spectra. With the help of the spectra, we then identify the causes of systematic deviations between lidar measurements and in situ anemometry. These are, first, spatial averaging along the measurement volumes and, second, cross-contamination between the three turbulence components, which we show has a strong influence on the shape of the spectra and the total variance of the measurement signal. Two methods are presented to improve lidar measurements of the longitudinal and vertical components of turbulence. First, we describe the method of squeezing that reduces the cross-contamination effect and can be applied to DBS and VAD scanning lidar. The method successfully reduces the effective separation distances between the line-of-sight measurement locations involved in the wind vector reconstruction. Second, we present a two-beam method that removes the spatial averaging along the measurement cone of VAD scanning lidars by only using the two lidar beams that point into the upstream and downstream directions. Floating lidar introduces additional challenges in accurately measuring turbulence, since the translational and rotational movement on water influences the measurement data. Using data collected on a VAD scanning lidar mounted on a floating buoy, we investigate the influence of motion in all six degrees of freedom on the wind measurements. We present a motion-compensation method that can correct for the motion-induced error on estimates of turbulence intensity, in cases when time series of motion data and line-of-sight velocities are available. This thesis concludes that turbulence measurements with currently available profiling wind lidars deviate significantly from one-point measurements. The data processing methods proposed here can overcome some of the measurement errors and can be implemented with existing lidars without changes to their hardware. Turbulence measurements from motion-compensated floating lidars can have an accuracy similar to measurements from fixed lidars. Overall, more work is needed to decrease the remaining uncertainty