Large scale atmospheric structures in space-time over flat terrain

From a wind energy perspective, large-scale coherent atmospheric structures are of particular interest when we want to understand the dynamics of downstream advection of wind turbine wakes and possible spatial variability in mean wind speed. The fast development of wind lidars and WindScanners makes now possible to measure them over large spatial domains, and understand better their characteristics under different atmospheric conditions.

This thesis presents the study of large-scale coherent structures in the atmospheric boundary layer over flat terrain, which was based on measurements of two scanning lidars deployed first, in the surface layer, and later in the mixed layer. This experiment, referred hereinafter as ”The Østerild Balconies Experiment”, was carried out during 2016 in Northern Jutland, Denmark.

The first part of this thesis deals with the processing of data from long-range pulsed lidars. It proposes a new methodology for data filtering and elaborates in what we can finally measure from this type of experiments. The proposed filter resulted in a substantial gain in availability and spatial coverage of good quality data. Regarding the velocity fluctuations we can measure, they appear to be limited to length scales larger than length of the sweeping step in the scanning pattern, which was around 250m in our case.

The second part shows the main findings from the experiment. For neutral atmospheric stability, within the surface layer, we identify streaks of alternated low and high momentum aligned with the mean wind speed. The flow structure appears horizontally homogeneous, but inhomogeneous in the vertical direction. Under unstable conditions the picture is very different, with frequent, large-scale convective rolls across the surface and mixed layer. Thermal convection makes the flow structure more homogeneous in the vertical direction, with convective rolls dominating the flow near the surface in a top-down process. Coherent structures under stable conditions are difficult to identify. Their are small scale nature is filtered during wind field reconstruction.

The final part of this thesis deals with the modelling of convective rolls as an extension of the uniform shear turbulence spectral tensor model. Using a continuous version of the discrete Fourier mode associated with rolls, the model is able to explain the large amplitudes in the low wavenumber range of the one-dimensional spectra. Moreover, with this formulation it is possible to include large-scale convective rolls in wind field simulations, allowing the fast generation of more realistic synthetic wind fields.