CISN ShakeAlert: Three Years of Comparative Real-Time Earthquake Early Warning Testing in California

The California Integrated Seismic Network (CISN) recently concluded a three-year project (August 2006-July 2009) aimed at the implementation, real-time testing, and comparative performance evaluation of three participating earthquake early warning (EEW) algorithms: 1) the Tau-C/P-d onsite algorithm developed by the California Institute of Technology, 2) the ElarmS algorithm developed by UC Berkeley, and 3) the Virtual Seismologist (VS) algorithm developed by the Swiss Seismological Service at ETH Zurich. These 3 EEW algorithms were installed and tested, and continue to run in real-time, at the Southern California Seismic Network, the Berkeley Digital Seismic Network, and the USGS Menlo Park network. The OnSite algorithm provides single-station magnitude estimates and estimates peak ground velocity at a given station. ElarmS and VS both provide magnitude and location estimates, as well as estimates of the geographic distribution of peak ground shaking. Over the last three years, these EEW algorithms submitted real-time and automatic non-interactive offline event reports to the Southern California Earthquake Center (SCEC) EEW Testing Center, which independently evaluated algorithm performance relative to the ANSS earthquake catalogue and observed ground motion datasets. We quantify the performance of these participating algorithms in terms of the accuracy of magnitude, location, and peak ground motion estimation, as well as the speed at which the algorithms provide information. Based on the derived performance characteristics, we infer how a prototype system based on the three algorithms might operate given alternative conditions, such as shorter telemetry delays, faster processing times, or higher station densities. The 2006-2009 CISN-EEW project demonstrated the feasibility and potential benefits of EEW in California. A new USGS-funded effort is underway to develop key components of CISN ShakeAlert, a prototype EEW system that could potentially be implemented in California. These key components include: 1) integration of the three algorithms into a single processing thread capable of providing alerts, 2) development of partnerships with key test users, and 3) development of protocols and procedures in collaboration with the test users that will govern the form and release of EEW information, if such a system is built.