In this study, we seek to bridge the global-scale weather regimes and the wind farm wake effects at the regional level, aiming to identify the influence of different circulation types on wake propagation. For this purpose, we use a two-step numerical approach. Initially, we use a mesoscale numerical weather prediction model to calculate the wind farm-produced wind speed deficits over three years, 2021, 2022, and 2023. Subsequently, we identify weather regimes using the ERA20C global reanalysis dataset from the ECMWF, covering a period from 1900 to 2010. A deep-learning-based dimensional reduction method is trained and validated on this long dataset, yielding a low-dimensional representation in which clusters corresponding to different weather regimes can be identified. The ERA5 dataset (1940-present) is additionally utilized to provide the weather regimes over the same period as the regional model. This exploratory study suggests that the different circulation types result in different wake patterns and varying wind farm power production in the Southern bight of the North Sea.