<p>Over the last years, installations of wind turbines (WTs) increased worldwide. Owing to<br>negative effects on humans, WTs are often installed in areas with low population density.<br>Because of low anthropogenic noise, these areas are also well suited for sites of<br>seismological stations. As a consequence, WTs are often installed in the same areas as<br>seismological stations. By comparing the noise in recorded data before and after<br>installation of WTs, seismologists noticed a substantial worsening of station quality leading<br>to conflicts between the operators of WTs and earthquake services.</p><p>In this study, we compare different techniques to reduce or eliminate the disturbing signal<br>from WTs at seismological stations. For this purpose, we selected a seismological station<br>that shows a significant correlation between the power spectral density and the hourly<br>windspeed measurements. Usually, spectral filtering is used to suppress noise in seismic<br>data processing. However, this approach is not effective when noise and signal have<br>overlapping frequency bands which is the case for WT noise. As a first method, we applied<br>the continuous wavelet transform (CWT) on our data to obtain a time-scale representation.<br>From this representation, we estimated a noise threshold function (Langston & Mousavi,<br>2019) either from noise before the theoretical P-arrival (pre-noise) or using a noise signal<br>from the past with similar ground velocity conditions at the surrounding WTs. Therefore, we<br>installed low cost seismometers at the surrounding WTs to find similar signals at each WT.<br>From these similar signals, we obtain a noise model at the seismological station, which is<br>used to estimate the threshold function. As a second method, we used a denoising<br>autoencoder (DAE) that learns mapping functions to distinguish between noise and signal<br>(Zhu et al., 2019).</p><p>In our tests, the threshold function performs well when the event is visible in the raw or<br>spectral filtered data, but it fails when WT noise dominates and the event is hidden. In<br>these cases, the DAE removes the WT noise from the data. However, the DAE must be<br>trained with typical noise samples and high signal-to-noise ratio events to distinguish<br>between signal and interfering noise. Using the threshold function and pre-noise can be<br>applied immediately on real-time data and has a low computational cost. Using a noise<br>model from our prerecorded database at the seismological station does not improve the<br>result and it is more time consuming to find similar ground velocity conditions at the<br>surrounding WTs.</p>