Knowing the wind speed across the rotor of a wind turbine is key in making good predictions of the power production. However, models struggle to capture both the speed and the shape of the wind profile. Using machine learning methods based on the model data, we show that the predictions can be improved drastically. The work focuses on three coastal sites, spread over the Northern Hemisphere (the Baltic Sea, the North Sea, and the US Atlantic coast) with similar results for all sites.

Low-level jets (LLJs) are special types of non-ideal wind profiles affecting both wind energy production and loads on a wind turbine. However, among LLJ researchers, there is no consensus regarding which definition to use to identify these profiles. In this work, we compare two different ways of identifying the LLJ – the falloff definition and the shear definition – and argue why the shear definition is better suited to wind energy applications.

Sometimes, the wind changes direction between the bottom and top part of a wind turbine. This affects both the power production and the loads on the turbine. In this study, a climatology of pronounced changes in wind direction across the rotor is created, focusing on Scandinavia. The weather conditions responsible for these changes in wind direction are investigated and the climatology is compared to measurements from two coastal sites, indicating an underestimation by the climatology.

As wind power becomes more popular, there is a growing demand for accurate power production forecasts. In this paper we investigated different methods to improve wind power forecasts for an offshore location in the Baltic Sea, using both simple and more advanced techniques. The performance of the methods is evaluated for different weather conditions. Smoothing the forecast was found to be the best method in general, but we recommend selecting which method to use based on the forecasted weather.

In this paper we study the impact of using wind observations from the Aeolus satellite, which provides wind speed profiles globally, in our numerical weather prediction system using a regional model covering the Nordic countries. The wind speed profiles from Aeolus are assimilated by the model, and we see that they have an impact on both the model analysis and forecast, though given the relatively few observations available the impact is often small.