Ocean-Bottom Seismograph Performance during the Cascadia Initiative

The intent of this analysis was to provide a summary of the characteristics and performance of the CI OBS instruments. The CI was the first community-driven experiment with OBSIP instrumentation and support and provided unprecedented geographic coverage due to the large number of available instruments and the four-year duration of the experiment. As a community experiment, CI data have been made immediately available to all members of the scientific community. The OBSIP facility has provided openly available horizontal orientation analysis and consistent metadata and conventions across all instrument types. The analysis of horizontal orientations was generally successful, though due to noise levels the Rayleigh-wave orientation methodology used here produced only a modest number of observations for each orientation estimate, and the error bars for these estimates are relatively large. A more comprehensive analysis of noise levels is warranted after the fourth year of data collection and will be the subject of future investigation when the data are recovered in September 2015. At that point there will be four years of data, and the sites in the northern and southern parts of the CI deployment area will have each been occupied twice. Noise-related questions that can be addressed more completely include the effect of instrument shielding on noise, and the behavior of noise as a function of both water depth and season. Analysis of seasonal effects in the CI data is particularly challenging, as the deployment and recovery of the instrumentation takes place predominantly during the summer months, thus interrupting the seasonal tracking of noise. A comparative analysis of the Cascadia OBS data to the Cascadia land-based stations could also lend insight into the characteristics of ocean-bottom seismic noise. Instrument improvement is an ongoing process, informed by prior year deployments. Making improvements in the CI instruments has been a challenge, given the minimal turnaround time between recovery and redeployment. Still, data recovery has steadily increased throughout the CI experiment. The CI experiment has an overall data recovery rate of 71% based on the 2011-2014 deployments and has increased the volume of broadband OBSIP data available in the IRIS DMC by ∼33%. The performance analysis can be used to inform the design of future AA experiments. The analysis of noise variables (e.g., water depth, seasonality, and shielding) could impact strategies for deployment locations and durations. For example, longer duration deployments of ∼15 months, which would span the summer months when deployment and/or recovery take place, could allow for improved analysis of the seasonality of noise during this experiment. The continuously improving performance in data return over the first three years of the CI experiment indicates a high probability of success in future experiments.