10+ Years of Ocean Bottom Seismometer Noise: Fresh Insights and Persistent Questions

Abstract:

The proliferation of broadband ocean bottom seismometer (OBS) deployments over the last two decades has generated key datasets from diverse marine environments, improving our understanding of tectonics and earthquake processes. In turn, the community of scientists using OBS data has also expanded, particularly with the emergence of several high-profile open-access experiments in the past decade, such as the Cascadia Initiative, Eastern North American Margin Community Seismic Experiment, and Alaska Amphibious Community Seismic Experiment. This growth in OBS data collection is likely to persist with the arrival of new seismic seafloor technologies, and continued scientific interest in marine and amphibious targets. The noise inherent in OBS data poses a challenge that is markedly different from that of terrestrial data. Use of this data and planning of future experiments benefits from an understanding of these challenges. However, the ways in which oceanographic noise affects OBS data across various marine deployment environments are not well understood, leading to limitations in our ability to prescribe best practices for utilizing this data as well as planning of future OBS experiments. Marine experiments increasingly comprise multiple instrument types, a trend that will increase as new technologies emerge. It is therefore essential to quantify noise and instrument behavior across the OBS pool and the range of deployment conditions. Using analysis of data from more than a decade of NSF-funded OBS deployments, we present systematically calculated tilt and compliance noise and corrections, and their relationship with instrument design, sensor type, water depth, and deployment location. We also find that ambient noise cross-correlations are affected by tilt and compliance noise, that processing techniques routinely used for terrestrial data, such as one-bit normalization, are less suitable in some oceanic environments, and that the extraction of high signal-to-noise Green’s functions may be more dependent on processing choices such as amplitude normalization than in terrestrial data. These findings will benefit the broader marine geophysics community by providing new recommendations for efficient use of OBS data.