<p>Ambient noise is a critical factor restricting the development of offshore seismic observation. Based on the power spectral density method, this paper systematically studies the ambient noise records at offshore platform-based seismic stations in the Bohai Sea area, and compares it with that at the coastal land-based station to obtain the characteristics and variation patterns of ambient noise across different frequency bands at offshore platform-based stations. The study shows that the ambient noise level at offshore platformbased stations is significantly higher than that at the land-based station and the new high noise model in the 0.1–2 Hz frequency band. Due to the special mechanical structure of the offshore platform, under the impact of waves, it will significantly amplify microseisms in the 0.1–2 Hz frequency band, particularly in the horizontal components; In the 2–20 Hz frequency band, the noise level at offshore platform-based stations is generally comparable to that at the land-based station, but there are peaks with the dominant frequency of around 5.5 Hz at the platform-based stations, which is speculated to originate from the mechanical vibration of the platforms; In the 0.01–0.1 Hz frequency band, the noise of horizontal components at offshore platform-based stations is higher than that at the land-based station, mainly due to long-period noise introduced by platform tilting caused by ocean waves and strong sea winds. There are two peaks that do not change with time in the noise power spectrum curves around 1 Hz, and we infer that they correspond to the natural frequencies of the offshore platform. The impact of sea winds on the ambient noise at offshore platform-based stations is mainly through energy transfer in the form of ocean waves, with the influence mainly concentrated in the frequency range of 0.1–2 Hz. Based on the probability density statistics of the continuous six-month ambient noise at offshore platform-based stations, the distribution of noise levels in each frequency band is generally stable over a long time scale. Affected by significant ambient noise, the original seismic records at offshore platform-based stations are unclear. By 0.2-1.5 Hz band-stop filtering, it is possible to maximize the suppression of high-level ambient noise while preserving as much information as possible about both local and teleseismic earthquakes. After filtering, the signal-to-noise ratio and seismic phase identification capability of both local and teleseismic earthquake records are significantly improved. This study provides theoretical basis for the large-scale deployment and data applications of offshore platform-based seismic stations in the future.</p>