Seismic noise characteristics of PNU OBS network in 2021-2022

Pusan National University deployed sixteen ocean bottom seismometers (OBSs) in the eastern offshore of the southern Korean peninsula. The primary purpose of the OBS network is monitoring earthquakes in the eastern offshore to investigate potential fault systems in the offshore region which was formidable using limited apertures by land-based observations. A seismic network's performance highly depends on each site's background noise level. We analyze the nature of the ambient noise and site response for ocean bottom seismometers (OBSs) deployed in the 2021-2022 period. The power spectral densities (PSDs) of the OBSs exhibited dis-similar features from those of land-based stations; most temporary broadband seismic stations on land showed relatively lower background noise levels. In the meanwhile, OBSs showed higher background noise levels. We report the nature of ambient noise at various channels, water depths, spatial locations, temporal variations, extreme weather conditions, and their potential causes. In general, horizontal components are noisier than vertical components. For longer periods, horizontal components are larger by ~45 dB. The probability density functions (PDFs) of OBSs show that the noise level is within the range of McNamara’s model (2004) for higher frequencies (3.5~50 Hz) although they are still high. When examining long periods (> 20 s), the noise level is higher than what would be given by McNamara’s model. Although we do not observe diurnal or weekly variations in OBS, as expected, we observe varying degrees of seasonal variations in OBSs. Apparently, water depth is the most important factor in deciding noise levels and their seasonal variations. At shallow-depth OBSs, we observe a strong correlation between noise levels and wave heights estimated by the Korea Meteorological Administration (KMA). The ambient noise is down to -130 dB for the band from 5 to 15 Hz, which provides the best signal-to-noise ratio for local microearthquakes. We also present the horizontal-to-vertical spectral ratio (HVSR) of the ambient noise recorded by OBSs. They present significant amplifications at lower frequencies, which indicates large amplification by combined effects due to lower density and lower wave velocity at shallow sediments, and greater depths to major impedance contrast. We confirm that modeling HVSR of noise data recorded by a three-component OBS offers a fast and inexpensive method for site investigation in deep water with the potential of in situ seafloor sediment characterization.