Abstract: The intensity of background noise is an important factor that affects the observation of seismic stations. Acquiring the characteristics of the background noise is of great significance to the evaluation of the quality of the data recorded by the Ocean Bottom Seismometer (OBS) and the noise reduction of the recorded data. Using the Probability Density Function (PDF) method to obtain the probability distribution characteristics of the Power Spectral Density (PSD) of the recorded data and comparing them with the results of new high-noise model (NHNM) and new low-noise model (NLNM) is a favorable method to research the background noise level around the station. Based on the long-term observation data of a large-scale passive source OBS array in the South China Sea (SCS), the background noise in the SCS is studied by using the PDF method. Firstly, the background noise in the whole frequency band is analyzed and compared with that from other stations. It is found that the background noise in the ocean is greater than the NHNM in the microseisms and low frequency band, and far greater than the background noise of the land-based station in the whole frequency band, which indicates that the data quality of the OBS is poor. Secondly, the probability density distribution of the earthquakes and other signals in the observation process is summarized, and it is found that the teleseismic, near earthquake and data dropout signal have different dominant frequency bands respectively, which has important significance for subsequent filtering processing and quality inspection. Finally, the time variation characteristics of background noise are studied, and it is found that typhoon is the main cause of time variation in microseisms period.