Self-noise models of five commercial strong-motion accelerometers

Abstract

Strong‐motion accelerometers provide onscale seismic recordings during moderate‐to‐large ground motions (e.g., up to tens of m/s 2 peak). Such instruments have played a fundamental role in improving our understanding of earthquake source physics (Bock etal. , 2011), earthquake engineering (Youd et al. , 2004), and regional seismology (Zollo et al. , 2010). Although strong‐motion accelerometers tend to have higher noise levels than high‐quality broadband velocity seismometers, their higher clip‐levels provide linear recordings at near‐field sites even for the largest of events where a collocated broadband sensor would no longer be able to provide onscale recordings (Clinton and Heaton, 2002). Recently, the seismological community has begun to make use of strong‐motion accelerometer data even in the absence of large ground motions (e.g., Tibuleac et al. , 2011). The noise floor of the instruments often limits the usefulness of strong‐motion accelerometer data in such studies, because it obscures first arrivals or can make the traces dominated by noise. When a strong‐motion accelerometer is deployed in a quiet setting, the noise floors of the digitizer and the accelerometer tend to dominate the other noise sources (Cauzzi and Clinton, 2013). This situation is unlike that using broadband sensors, in which site conditions are typically the largest contributing source of noise in seismic data, especially at long periods (Wilson et al. , 2002). With the widespread deployment of strong‐motion accelerometers recorded on high resolution digitizers, it is now possible to get continuous high‐rate acceleration data in which the digitizer noise is not the dominant noise source (Cauzzi and Clinton, 2013). To better characterize the noise of a number of commonly deployed accelerometers in a standardized way, we conducted noise measurements on five different models of strong‐motion accelerometers. Our study was limited to traditional accelerometers (Fig. 1) and is in no way exhaustive.