Solar radiation has a fluctuating nature, especially in oceanic climates where cloud-induced fluctuations are very common. The fluctuations and thus the power system reserve requirements can differ a lot on a daily and hourly basis. In this paper, local hourly classification criteria are defined based upon a spectral analysis of the wavelet transform by using 10 sec solar irradiation data. Three distinctly different classes are identified: overcast, partly cloudy and sunny. The classification is extended to the national scale for 8 sites throughout the Netherlands using 10 min data. Although the nationwide characteristics cannot be described by one single class, the classification still remains meaningful as the fraction of locations with partly cloudy conditions determines the level of fluctuations. Evaluation of the results show the impact of the classification and the consequences for the required system reserves. A combination of the classification method with dynamic reserve allocation is proposed, which should contribute to a more efficient and reliable power system operation.@en