Data Delivery Latency Improvements And First Steps Towards The Distributed Computing Of The Caltech/USGS Southern California Seismic Network Earthquake Early Warning System

The USGS/Caltech Southern California Seismic Network (SCSN) is a modern digital ground motion seismic network. It develops and maintains Earthquake Early Warning (EEW) data collection and delivery systems in southern California as well as real-time EEW algorithms. Recently, Behr et al., SRL, 2016 analyzed data from several regional seismic networks deployed around the globe. They showed that the SCSN was the network with the smallest data communication delays or latency. Since then, we have reduced further the telemetry delays for many of the 330 current sites. The latency has been reduced on average from 2-6 sec to 0.4 seconds by tuning the datalogger parameters and/or deploying software upgrades. Recognizing the latency data as one of the crucial parameters in EEW, we have started archiving the per-packet latencies in mseed format for all the participating sites in a similar way it is traditionally done for the seismic waveform data. The archived latency values enable us to understand and document long-term changes in performance of the telemetry links. We can also retroactively investigate how latent the waveform data were during a specific event or during a specific time period. In addition the near-real time latency values are useful for monitoring and displaying the real-time station latency, in particular to compare different telemetry technologies. A future step to reduce the latency is to deploy the algorithms on the dataloggers at the seismic stations and transmit either the final solutions or intermediate parameters to a central processing center. To implement this approach, we are developing a stand-alone version of the OnSite algorithm to run on the dataloggers in the field. This will increase the resiliency of the SCSN to potential telemetry restrictions in the immediate aftermath of a large earthquake, either by allowing local alarming by the single station, or permitting transmission of lightweight parametric information rather than continuous waveform data to the central processing facility. State-of-the-art development of Internet of Things (IoT) tools and platforms, which can be used to distribute and maintain software on a large number of remote devices are making this approach to earthquake early warning more feasible.