Converting strong-motion networks to arrays via common triggering

The ICEARRAY network of accelerographs was deployed in the south Iceland seismic zone in the fall of 2007. It consists of 14 stations each equipped with the CUSP-3Clp stand-alone triaxial accelerographs running in triggered mode, configured with a wireless and perpetual GPRS communication link and a GPS timing system. The array is of small aperture (1.9 km) over which a relatively high waveform correlation has been observed (Halldórsson et al. 2009). However, the ICEARRAY is located in an urban environment where natural and cultural conditions lead to different and variable background noise levels at different stations. As a result, the network generally fails to capture earthquake weak motion on all array stations, resulting in incomplete data sets. Moreover, the large number of noise-related data files produced by individual array stations results in considerable effort being wasted in manually reviewing useless data. This has led to the development of a CT scheme, in which the instruments are configured to send trigger notification messages to one or more central hubs running the CT algorithm. Each trigger notification message is decoded to find its source, and on this basis a preset number of votes are added to a tally. Whenever a specific number of votes has accumulated within a given moving time window, a global triggering command is issued to all instruments within the network, producing a complete data set. Thus, the CT scheme operating over the Internet has maximized the efficiency of the triggering system of the ICEARRAY, effectively converting the network into a triggered array that records significant (real) events only. In general, the CT scheme markedly improves the usefulness of a network of stand-alone instruments and effectively eliminates the need for manually reviewing recorded data to separate earthquake recordings from noise. Additionally, it allows sites with marginal triggering suitability to be effectively incorporated as slave instruments. While this may especially be the case for small-aperture networks of strong-motion instruments, it applies also to large-aperture strong-motion networks. In the latter case however, in order to avoid triggering distant instruments where the ground motion is below the noise floor, the CT scheme simply triggers the subgroup of instruments where it is ensured that actual and useful waveforms can be recorded. For this purpose, the CT scheme would simply require information on the regional attenuation of earthquake ground motion. Finally, the authors hope that the packet structure of the CT scheme can form the basis of an open and manufacturer-independent protocol that allows a diverse range of instruments and sensor types to be incorporated.