A Method to Establish Seismic Noise Baselines for Automated Station Assessment

Abstract

Currently, the United States Geological Survey (USGS) National Earthquake Information Center (NEIC) receives more than 4,000 channels of seismic data from roughly 900 seismic stations distributed around the world in near real time. As the NEIC develops increasingly sophisticated methods for automated data processing, it becomes equally important to develop tools for ensuring that only the highest-quality data are used in rapid earthquake products. In this paper, we present methods and applications for computing long- and short-term, station-specific noise baselines for broadband seismic data channels. Long-term baselines (one year or greater) are useful for determining ambient noise conditions, while short-term baselines (one hour to one month) are useful for monitoring changing station performance and noise characteristics. This method is currently in testing and development at the NEIC for evaluation of instrument responses, assessment of data quality for real-time earthquake processing systems, new station site evaluation, and assessment of gross network characteristics. Real-time earthquake processing systems at the NEIC rely on high-quality seismic data to compute accurate earthquake locations and magnitudes, moment-tensor solutions, finite-fault models, and shaking intensity. The NEIC receives real-time seismic data from a variety of contributors including: global networks (Global Seismographic Network [GSN], GEOSCOPE, GEOFON), national networks (ANSS, Australia, Spain, Canada, Switzerland), regional networks (ANZA, PRSN, CISN, PNWSN, AEIC), government agencies (National Oceanic and Atmospheric Administration [NOAA] tsunami warning centers, the International Monitoring System [IMS]), and temporary deployments such as the EarthScope transportable array (TA). On most days, approximately 10% of contributed instrument responses are either not known or are erroneous and cannot be used for accurate real-time amplitude information. Maintaining accurate and current metadata for roughly 900 contributed stations is difficult when instrumentation upgrades or other changes occur regularly. Automated processing of real-time seismic data necessitates the development of automated tools and procedures to quickly identify changes …