Mesoscale wind fluctuations over Danish waters

Mesoscale wind uctuations aect the large scale integration of wind power because they undermine the day-ahead predictability of wind speed and power production, and because they can result in large uctuations in power generation that must be balanced using reserve power. Large uctuations in generated power are a particular problem for oshore wind farms because the typically high concentration of turbines within a limited geographical area means that uctuations can be correlated across large numbers of turbines. Furthermore, organised mesoscale structures that often form over water, such as convective rolls and cellular convection, have length scales of tens of kilometers, and can cause large wind uctuations on a time scale of around an hour. This thesis is an exploration of the predictability of mesoscale wind uctuations using observations from the world's rst two large oshore wind farms - Horns Rev I in the North Sea, and Nysted in the Baltic Sea. The thesis begins with a climatological analysis of wind uctuations on time scales of 1{10 hours at the two sites. A novel method for calculating conditional climatologies of spectral information is proposed, based on binning and averaging the time axis of the Hilbert spectrum. Results reveal clear patterns between wind uctuations and locally observed meteorological conditions. The analysis is expanded by classifying wind uctuations on time scales of 1{3 hours according to synoptic patterns, satellite pictures and wind classes. Results indicate that cold air outbreaks and open cellular convection are a signicant contributor to mesoscale wind variability at Horns Rev. The predictability of mesoscale wind uctuations is tested by implementing standard statistical models that relate local wind variability to parameters based on a large scale weather analysis. The models show some skill, but only achieve a 15% improvement on a persistence forecast. The possibility of explicitly modelling mesoscale uctuations in a mesoscale model is then examined using the weather research and forecasting (WRF) model. A set of case studies demonstrate that realistic hour-scale wind uctuations and open cellular convection patterns develop in WRF simulations with 2km horizontal grid spacing. The atmospheric conditions during one of the case studies are then used to initialise a simplied version of the model that has no large scale weather forcing, topography or surface inhomogeneties. Using the simplied model, the sensitivity of the modelled open cellular convection to choices in model setup and to aspects of the environmental forcing are tested. Finally, the cell-scale kinetic energy budget of the modelled cells is calculated, and it is shown that the buoyancy and pressure balance terms are important for cell maintenance. It is explained that the representation of mesoscale convection in a mesoscale model is not only important to end users such as wind farm operators, but to the treatment of energy transport within the boundary layer.